Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[opticyclic]

Firstly, why is there no Bug Tracker such as JIRA for the git project?
This mailing list is next to useless for users since searching is
difficult, as is commenting and voting.

Secondly, since one of the alleged reasons for creating git was to not
have to deal with patches, why are pull requests disable and patches
sent to this mailing list?!
I have read https://github.com/gitster/git/blob/master/Documentation/SubmittingPatches
and it doesn't explain it.

I'm sure I don't have to tell you that GitHub has discussions on pull
requests, which are easier to view than the mailing list archives.

So why is it done in the current way?

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Brian Gernhardt]

On Feb 29, 2012, at 12:19 PM, opticyclic wrote:

> Secondly, since one of the alleged reasons for creating git was to not
> have to deal with patches, why are pull requests disable and patches
> sent to this mailing 


On the contrary, the design of git was created with the idea of handling patches via e-mail in mind.  Tools like git-format-patch, git-am, and git-send-email exist to allow this workflow.  Linus wanted a tool that would automate and enhance the way the kernel ML already did work instead of demanding that they change.

~~ Brian G

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

opticyclic <opticyclic@gmail.com> writes:

> Firstly, why is there no Bug Tracker such as JIRA for the git project?

Probably because nobody volunteered to set-up, actively de-dupe, triage
and maintain it in general.

> Secondly, since one of the alleged reasons for creating git was to not
> have to deal with patches, why are pull requests disable and patches
> sent to this mailing list?!

I think Brian already corrected whoever "alleges" such.

We prefer to develop in the open, reviewing and improving both patches and
ideas on the mailing list, without having to rely on a single project
hosting site everybody has to go and deal with web based interface.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Jonathan Nieder]

Junio C Hamano wrote:
> opticyclic <opticyclic@gmail.com> writes:

>> Firstly, why is there no Bug Tracker such as JIRA for the git project?
>
> Probably because nobody volunteered to set-up, actively de-dupe, triage
> and maintain it in general.

By the way, my usual offer/shameless plug[*] still stands: anyone who
can stand the interface is welcome to file, triage, and work on bugs
in the bugtracker at <http://bugs.debian.org/src:git>, as long as it
seems possible that your bugs might also affect Debian.

"Work on" usually means "forward to the git mailing list", but maybe
having a bug number is a comfort to some people. ;-) See
<http://www.debian.org/Bugs/Reporting> for instructions.

All that said, that is still not The Bug Tracker for the git project.
I would not want it advertised on git-scm.com until we have had some
more practice dealing with outside bugs, and maybe more contributors
sorting through them.

It may be that others provide a similar service.

Hope that clarifies a little,
Jonathan

[*] http://thread.gmane.org/gmane.comp.version-control.git/181336/focus=181402

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 09:53, Jonathan Nieder <jrnieder@gmail.com> wrote:
> Junio C Hamano wrote:
>> opticyclic <opticyclic@gmail.com> writes:
>
>>> Firstly, why is there no Bug Tracker such as JIRA for the git project?
>>
>> Probably because nobody volunteered to set-up, actively de-dupe, triage
>> and maintain it in general.
>
> By the way, my usual offer/shameless plug[*] still stands: anyone who
> can stand the interface is welcome to file, triage, and work on bugs
> in the bugtracker at <http://bugs.debian.org/src:git>, as long as it
> seems possible that your bugs might also affect Debian.
>
> "Work on" usually means "forward to the git mailing list", but maybe
> having a bug number is a comfort to some people. ;-) See
> <http://www.debian.org/Bugs/Reporting> for instructions.
>
> All that said, that is still not The Bug Tracker for the git project.
> I would not want it advertised on git-scm.com until we have had some
> more practice dealing with outside bugs, and maybe more contributors
> sorting through them.
>
> It may be that others provide a similar service.
>
> Hope that clarifies a little,
> Jonathan
>
> [*] http://thread.gmane.org/gmane.comp.version-control.git/181336/focus=181402

I mentioned I was going to do this a while ago, but decided to bite
the bullet and actually do it.

I have set up a JIRA instance using Atlassian's OnDemand service,
available at https://git-scm.atlassian.net/

The set-up is a work in progress, so don't be surprised if you log in
and it is not brimming with content!

For now I have locked it down so that anyone can sign up and log in,
but only approved users can create issues. I imagine this restriction
will be loosened down the track, but for now if you would like to
contribute please email me directly with your username and I will add
you as a 'Trusted User' (you will need to sign up first). For now,
normal users and trusted users are identical except that trusted can
create new issues.
If you want to update issues you will need to be a 'Developer' - if
you think you fall into this category please tell me and I will add
you to it as well (this is the role I think most people active on this
list will need to be)! This will allow you to edit issues, be assigned
issues and resolve them.

If anyone has experience administrating JIRA and want to help out
_please_ let me know as I know how quickly these things can grow and
become unmanageable. Additionally, if anyone has any ideas, or thinks
we should throw the gates open now (as opposed to locking it down for
a time) comment here and we can make it happen.

As I see it (and Junio has mentioned before) we are going to need
people who are able to manage the issues in this system, ensuring that
stale issues are closed out and progress matches what is happening on
the list. There is no intention to replace the list as a place where
issues are reported and discussed, however that may happen to some
degree anyhow. We should consider how a dedicated bug tracker like
this might impact the community and design for that.
For my mind, if JIRA was to become a 'mirror' of what goes on here in
the list that is ok - perhaps a duplication of efforts, but ok. If
nothing else, we will have a structured store of information keeping
track of issues that is very easy to access and work with.

I don't think I have enough context on all that is going on on the
list to be able to keep everything up-to-date myself, so if you think
you can assist in this capacity, again please contact me.

In case you were interested, some information on me:
I am a keen git user of who-knows-how-many years, and have been
following the list for the last 12 months or so. My day job is as a
JIRA Consultant (and everything else!) with the premier Atlassian
services partner [1].

Regards,

Andrew Ardill

[1] customware.net - we also work with Zendesk, and offer custom
plugins, customisations, theming, training and support packages as
well

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Greg Troxel]

[-- Attachment #1: Type: text/plain, Size: 224 bytes --]


  I have set up a JIRA instance using Atlassian's OnDemand service,
  available at https://git-scm.atlassian.net/

Do people really think it's reasonable to use non-Free tools to develop
git?  That seems surprising to me.


[-- Attachment #2: Type: application/pgp-signature, Size: 194 bytes --]

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 11:37, Greg Troxel <gdt@ir.bbn.com> wrote:
>
>  I have set up a JIRA instance using Atlassian's OnDemand service,
>  available at https://git-scm.atlassian.net/
>
> Do people really think it's reasonable to use non-Free tools to develop
> git?  That seems surprising to me.
>

Maybe not, and if that is the case I am more than happy to let this die.

That said, this is the tool I know how to use best, and is in my
opinion the most flexible, reliable and supported. The source code is
available on request to customers to extend or modify (or at least it
used to be) and the company is very supportive of open source projects
in general.

Additionally, if we are not prepared to use non-Free tools, we should
probably stop using github. (This example is a little trite, seeing as
there are non-github alternatives available for grabbing the source
code. Then again, the mailing list is not disappearing any time soon,
so there is a free alternative to _any_ bug tracker that is used)

Regards,

Andrew Ardill

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Andrew Ardill <andrew.ardill@gmail.com> writes:

> Additionally, if we are not prepared to use non-Free tools, we should
> probably stop using github. (This example is a little trite, seeing as
> there are non-github alternatives available for grabbing the source
> code.

Just on this part.

Github is not the only place to grab the source code.  Far from it.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 11:50, Junio C Hamano <gitster@pobox.com> wrote:
> Andrew Ardill <andrew.ardill@gmail.com> writes:
>
>> Additionally, if we are not prepared to use non-Free tools, we should
>> probably stop using github. (This example is a little trite, seeing as
>> there are non-github alternatives available for grabbing the source
>> code.
>
> Just on this part.
>
> Github is not the only place to grab the source code.  Far from it.

I understand that Github is not the only place to grab the source code
(maybe it was not clear that I understood that), however the point was
more that even though Github is not-Free many people still use it to
develop Free software (including people developing git). As Free
alternatives to Github are available not many people mind too much (or
so it seems).

Why do people use Github at all? Perhaps, because it provides an
accessible, reliable, powerful and supported platform, with large
amount of penetration in the market and some very desirable features.

I believe that JIRA in the OnDemand package offers similar benefits.
Additionally, Free alternatives would still be available (the mailing
list). Perhaps there is too much controversy to anoint a JIRA issue
tracker as 'official', however I continue to hear people ask for a
tracker, and apart from Jonathan Nieder with the Debian bug tracker
see no one else putting their hands up.

In any case, I spun the instance up because nothing happens until
someone does something, and if it fails then at least we have a record
of trying it next time someone asks :)

I would love to see it succeed.

Regards,

Andrew Ardill

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Andrew Ardill <andrew.ardill@gmail.com> writes:

> On 1 March 2012 11:37, Greg Troxel <gdt@ir.bbn.com> wrote:
>>
>> Do people really think it's reasonable to use non-Free tools to develop
>> git? That seems surprising to me.
>
> Maybe not, and if that is the case I am more than happy to let this die.

I won't speculate how big or small part of the Git community you would be
repelling by using a closed/commertial offering. It may not be such a big
deal, or it may be. I simply do not know.

But we will never find out until we try. The same thing can be said for
the usefulness of having a bug tracker and feasibility of keeping it
reasonably clean and useful over time with volunteer effort.  I commend
you for finally stepping up and biting the bullet to start an experiment.

One request I may have is to give read/browse-only access to unregistered
users without any account (I hate having to maintain credentials to random
websites, and I imagine so do many other people), but I am not the target
audience, so please do not bend backwards to implement such if it is too
much trouble with the system.

Thanks.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 12:05, Junio C Hamano <gitster@pobox.com> wrote:
> Andrew Ardill <andrew.ardill@gmail.com> writes:
>
>> On 1 March 2012 11:37, Greg Troxel <gdt@ir.bbn.com> wrote:
>>>
>>> Do people really think it's reasonable to use non-Free tools to develop
>>> git? That seems surprising to me.
>>
>> Maybe not, and if that is the case I am more than happy to let this die.
>
> I won't speculate how big or small part of the Git community you would be
> repelling by using a closed/commertial offering. It may not be such a big
> deal, or it may be. I simply do not know.

Neither do I - let's find out!

> But we will never find out until we try. The same thing can be said for
> the usefulness of having a bug tracker and feasibility of keeping it
> reasonably clean and useful over time with volunteer effort.  I commend
> you for finally stepping up and biting the bullet to start an experiment.

I have been meaning to get this going for a few months now, the latest
thread kickstarted me again.

> One request I may have is to give read/browse-only access to unregistered
> users without any account (I hate having to maintain credentials to random
> websites, and I imagine so do many other people), but I am not the target
> audience, so please do not bend backwards to implement such if it is too
> much trouble with the system.
>
> Thanks.

I have given browse access by default to Anyone (does not require
log-in). JIRA is highly customisable, so most requests are typically
feasible.

Regards,

Andrew Ardill

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Miles Bader]

Andrew Ardill <andrew.ardill@gmail.com> writes:
> Additionally, if we are not prepared to use non-Free tools, we should
> probably stop using github.

I think the issue isn't "using" non-free tools so much as it's getting
_locked into_ non-free tools.

So if JIRA makes all its data trivially exportable in a format which
is easy to use, and there's some way of isolating references to it so
that it's possible to switch to something else without undue hardship,
maybe it's not such a big deal.

[Github is a prime example of a non-free tool which is largely avoids
the lockin issue, and I imagine that's why so many free projects
happily use it.  That's generally true of any git repo site, given the
super-easy cloneability of git repos, but Github also provides good
export of the meta-data it keeps (issues, etc).]

-Miles

-- 
/\ /\
(^.^)
(")")
*This is the cute kitty virus, please copy this into your sig so it can spread.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 16:16, Miles Bader <miles@gnu.org> wrote:
> Andrew Ardill <andrew.ardill@gmail.com> writes:
>> Additionally, if we are not prepared to use non-Free tools, we should
>> probably stop using github.
>
> I think the issue isn't "using" non-free tools so much as it's getting
> _locked into_ non-free tools.
>
> So if JIRA makes all its data trivially exportable in a format which
> is easy to use,

JIRA provides the ability to trivially export any 'filter' as XML or
CSV data files. A filter is simply a search, and can contain every
single issue in the database. Some pieces of information are slightly
more difficult to migrate.
- Issue attachments such as screenshots are not stored in the
database, and we would have to request them in order to retrieve them.
- User information and configuration data is only available in a full
site backup (which is available on request).
- CSV export option does not include comments. (XML, and Word formats do)

Additionally, there are quite a number of tools to allow migration
between JIRA and other tools, and in my experience Atlassian are quite
helpful when you try and move off their products.

> and there's some way of isolating references to it so
> that it's possible to switch to something else without undue hardship

JIRA has the capacity to automatically link issues to commits which
mention issue numbers, however this is by no means essential or
necessary functionality. JIRA will be happy completely detached from
the rest of the world, and will provide additional functionality
gradually as it becomes more integrated. Some functionality it can
provide without forming hard links with the outside world (like
polling the repository and providing continuous builds)

The primary thing to look out for would be other sites linking to the
issue tracker. Unfortunately, there is no way at the moment to provide
an alternate URI for the tracker, so it is hard to avoid this from
happening.

Regards,

Andrew Ardill

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Scott Chacon]

On Wed, Feb 29, 2012 at 4:37 PM, Greg Troxel <gdt@ir.bbn.com> wrote:
>
>  I have set up a JIRA instance using Atlassian's OnDemand service,
>  available at https://git-scm.atlassian.net/
>

Honestly I would argue against this, just because unless you want to
spend a lot of time on it, I don't think it's going to get used much.
Issue trackers in general tend not to get traction unless the
maintainer uses it and asks people to use it or it's included in the
workflow somehow.  I use issue trackers in most of my projects, but I
also don't use mailing lists - a lot of the things that work for my
workflows are not the way that the Git project does it and I think
you'll find that it's a bit of a waste of time to try to shoehorn them
in.

Besides, most of the things you are looking to get out of this are
generally pretty easily obtained from the ML.  If you're bored and
want a project to work on, ask the ML.  If you want to know the status
on something, search or ask the ML.  It's not quite as self-service as
a issue tracker, but it gets you into the community more, which I
think is also important.

> Do people really think it's reasonable to use non-Free tools to develop
> git?  That seems surprising to me.

This is particularly interesting to me.  Disclaimer: I work at GitHub
and have for most of the life of GitHub.  That said, it's interesting
to think about this.  What does the freedom of the tooling provide
you?  Data portability is one thing, but both JIRA and GitHub have
APIs to obtain basically any data in them (I think - I never use JIRA,
but I've worked on the GH APIs).

I do think that an interesting data point here is the cast of
kernel.org, though.  So that's all free, but also hugely and totally
failed everyone here.  It wasted hours of my time trying to clean up
all the broken links from git-scm.com over a month, which is after a
month of thinking that they couldn't possibly be down another day.
The wiki is still busted.  The docs are still gone.  GitHub has
contracted a designer and has started spending developer hours working
on a better git-scm.com to take over those functions so it won't
happen again.  More importantly, that will never happen to GitHub or
JIRA - there is no conceivable way that either of these relatively
large corporations would tolerate even a full day of downtime or data
loss.

So free is great, but what is more important in the tooling and
services that help you develop?  Is it freedom to some arbitrary
level, or is it simplicity and availability? I value my time a lot
more than if I can get the source code to the issue tracker that my
open source project uses.  If we're going to use an issue tracker, or
any other tool, I would really rather prefer we use one backed by a
company that takes downtime seriously as opposed to using something
that doesn't have the resources to fix things in any timeframe.
Having someone saying "it's going to keep working because if it
doesn't we all lose our jobs" is more freedom to me than having people
say "if it doesn't work at some point you have the freedom to spend
days of your time reimplementing it on your own hardware with maybe
some of our backed up data and our open source code".  Which is
literally what I'm doing today for the hosted man page documentation.

Just some thoughts.

Scott

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Scott Chacon <schacon@gmail.com> writes:

> So free is great, but what is more important in the tooling and
> services that help you develop?  Is it freedom to some arbitrary
> level, or is it simplicity and availability? I value my time a lot
> more than if I can get the source code to the issue tracker that my
> open source project uses.

I do not particularly want to say this, but I couldn't resist wondering
how the above "The best tool for the job, be it libre or commercial"
compares with what Linus would have had in his mind when the kernel
project decided to use BitKeeper.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Thomas Rast]

Andrew Ardill <andrew.ardill@gmail.com> writes:

> I have set up a JIRA instance using Atlassian's OnDemand service,
> available at https://git-scm.atlassian.net/
[...]
> As I see it (and Junio has mentioned before) we are going to need
> people who are able to manage the issues in this system

Note that you are not the first one to try.  The most elaborate plan and
writeup that I know of sits at

  http://article.gmane.org/gmane.comp.version-control.git/136500  [1]

Jan "jast" Krüger also mentioned server issues today, so *.jk.gs is
presumably down because of that, not because gitbugs.jk.gs is no longer
valid.

Nevertheless, AFAIK it has never been used for "real work", so you may
want to look into why that happened, and do something different.

-- 
Thomas Rast
trast@{inf,student}.ethz.ch

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Ævar Arnfjörð Bjarmason]

On Thu, Mar 1, 2012 at 12:29, Thomas Rast <trast@inf.ethz.ch> wrote:
> Andrew Ardill <andrew.ardill@gmail.com> writes:
>
>> I have set up a JIRA instance using Atlassian's OnDemand service,
>> available at https://git-scm.atlassian.net/
> [...]
>> As I see it (and Junio has mentioned before) we are going to need
>> people who are able to manage the issues in this system
>
> Note that you are not the first one to try.  The most elaborate plan and
> writeup that I know of sits at
>
>  http://article.gmane.org/gmane.comp.version-control.git/136500  [1]
>
> Jan "jast" Krüger also mentioned server issues today, so *.jk.gs is
> presumably down because of that, not because gitbugs.jk.gs is no longer
> valid.
>
> Nevertheless, AFAIK it has never been used for "real work", so you may
> want to look into why that happened, and do something different.

As someone who submits patches every once in a while I can echo other
sentiments in this thread, just because you have a list of issues that
doesn't mean anyone is working on them.

However I'd also sometimes like to work on some random issue because
I'm bored, and having a collection of issues ordered by priority (or
popularity) would be useful when that happens.

But I think any proposal to set up a wholly external system is going
to fail, we do most of our bug submission / commenting etc. on this
mailing list, and that isn't going to change, so there's always going
to be a large chasm between the list and any external system.

What I think *would* work however is a system that feeds off the
mailing list. This could be as simple as a mailing list aggregator
that allowed you to star certain messages, and the most starred
messages would be the popular issues.

A more fancy solution would:

 * Consume every single message that gets sent to the list
 * Group each thread and allow it to be categorized as a
   bug/issue/enhancement/complaint
 * Allow you to mark a collection of threads as describing the same
   issue, so you'd have duplicates marked & the full history of a
   discussion on some issue.
 * Allow you to mark an issue as outstanding / resolved / allow voting
   on it.

Thus you'd automatically build up an issue database without anyone
going out of their way, all it would need is the same people who
complain that they can't file bugs either categorizing existing posts,
or categorizing a post they just made.

Many bug trackers can be made to work with E-Mail (e.g. Jira, RT
etc.), although I don't know if they're well set up to follow a
mailing list like this. I think e.g. Jira assumes that you have a the
bug id in the subject, and might not be smart enough to group things
by In-Reply-To headers.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 22:54, Ævar Arnfjörð Bjarmason <avarab@gmail.com> wrote:
> As someone who submits patches every once in a while I can echo other
> sentiments in this thread, just because you have a list of issues that
> doesn't mean anyone is working on them.
>
> However I'd also sometimes like to work on some random issue because
> I'm bored, and having a collection of issues ordered by priority (or
> popularity) would be useful when that happens.

JIRA is particularly good at categorising backlogs and filtering out
issues, so this is a very achievable outcome.

> But I think any proposal to set up a wholly external system is going
> to fail, we do most of our bug submission / commenting etc. on this
> mailing list, and that isn't going to change, so there's always going
> to be a large chasm between the list and any external system.

I agree wholeheartedly.

> What I think *would* work however is a system that feeds off the
> mailing list. This could be as simple as a mailing list aggregator
> that allowed you to star certain messages, and the most starred
> messages would be the popular issues.

This is an interesting idea, not sure how that would be implemented
though. Seems like you would need some client or webservice email
reader that could be extended like that.

> A more fancy solution would:
>
>  * Consume every single message that gets sent to the list
>  * Group each thread and allow it to be categorized as a
>   bug/issue/enhancement/complaint
>  * Allow you to mark a collection of threads as describing the same
>   issue, so you'd have duplicates marked & the full history of a
>   discussion on some issue.
>  * Allow you to mark an issue as outstanding / resolved / allow voting
>   on it.
>
> Thus you'd automatically build up an issue database without anyone
> going out of their way, all it would need is the same people who
> complain that they can't file bugs either categorizing existing posts,
> or categorizing a post they just made.
>
> Many bug trackers can be made to work with E-Mail (e.g. Jira, RT
> etc.), although I don't know if they're well set up to follow a
> mailing list like this. I think e.g. Jira assumes that you have a the
> bug id in the subject, and might not be smart enough to group things
> by In-Reply-To headers.

JIRA allows creation of issues by sending emails to a special email
address. The process that is used is as follows (take from [1]):

The subject  of an email message is examined for an existing issue key:
 - If an issue key is found in the subject, the content of the email
message's body is processed and added as a comment to the issue with
that issue key.
 - If an issue key is NOT found in the subject, the in-reply-to header
 is examined:
    - If the email message is found to be a reply to another email
message from which an issue was previously created, the body is
processed and added as a comment to that issue.
    - If the email message is NOT found to be a reply, a new issue is created.

As you can see, it _should_ respect in-reply-to headers, however we
should probably test this :)

Currently only emails that contain issue keys are being captured,
however I will enable issue creation in a throwaway project so we can
test them. I might even set up a forwarder to capture all or most of
the list traffic to see what happens. If we capture any useful issues
in that throwaway project, a valid workflow might be to move them over
to the 'real' project to which they belong. Perhaps the volume will be
low enough in reality that we can enable issues being created directly
in that project, without the move step.

One thing to consider is the notifications sent out by JIRA on
different events. We could potentially send an email to the list
whenever an issue is commented on, resolved, or something else. The
possibilities (and permissions around those possibilities) are quite
versatile, and well worth investigating. For now I will try my best to
stop _all_ automated responses going to the list, until we can be
certain that it won't be just more spam.


Thanks for your thoughts, I think your ideas give a really strong
direction for us to investigate further.

Regards,

Andrew Ardill

[1] http://confluence.atlassian.com/display/JIRA/Creating+Issues+and+Comments+from+Email#CreatingIssuesandCommentsfromEmail-Issuecommentcreation

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

On 1 March 2012 22:29, Thomas Rast <trast@inf.ethz.ch> wrote:
> Andrew Ardill <andrew.ardill@gmail.com> writes:
>
>> I have set up a JIRA instance using Atlassian's OnDemand service,
>> available at https://git-scm.atlassian.net/
> [...]
>> As I see it (and Junio has mentioned before) we are going to need
>> people who are able to manage the issues in this system
>
> Note that you are not the first one to try.  The most elaborate plan and
> writeup that I know of sits at
>
>  http://article.gmane.org/gmane.comp.version-control.git/136500  [1]
>
> Jan "jast" Krüger also mentioned server issues today, so *.jk.gs is
> presumably down because of that, not because gitbugs.jk.gs is no longer
> valid.
>
> Nevertheless, AFAIK it has never been used for "real work", so you may
> want to look into why that happened, and do something different.
>

Thanks for the links Thomas, I think the general sentiments summarised
there are echoed every time someone mentions an issue tracker on the
list. It is good to have the summary though!

I think that, to some degree, the model that most quickly arises from
the ideals of that list is inherently flawed. The git list is, almost
by definition, a fairly chaotic place. Issues get looked into because
either someone makes a loud enough noise, or someone is interested
enough in the problem. An issue tracker is a much more structured
beast, and trying to tether the two is obviously going to be
difficult.
One of the benefits of an issue tracker is that it keeps all
discussion around an issue in the one place. The driving wedge here is
that we *already* have somewhere that issues are tracked, albeit in a
less structured sense. Anything that reduces the duplication efforts
needed is a desirable thing, but more on that later.

The one thing we need to avoid is single points of failure. In my
experience, this is achieved by ensuring that people contributing
issues are able to do as much of the leg work as possible. Typically
they want to help, and are limited most by unnecessary restrictions
and lack of understanding. If the process is simple enough it becomes
much easier for new people to step in and make valuable contributions
(the mailing list is in some ways the embodiment of this). We also
need to encourage those who are willing and empower each other to be
time-effective.

I feel that is enough pontificating from me, and I am really enjoying
the discussion. Please, if you think it will work or not, visit the
site [1] - sign up if you want to - and provide feedback. I am going
to open a project to capture feedback, and anything else related that
is not directly relevant to the list, and that will probably be a good
place to test the ropes if you want to get your hands dirty. In
general, it is very hard to break things so if you would like to have
a play around with the system please let me know and get involved!

Regards,

Andrew Ardill

[1] git-scm.atlassian.net

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Thomas Rast <trast@inf.ethz.ch> writes:

> Nevertheless, AFAIK it has never been used for "real work", so you may
> want to look into why that happened, and do something different.

Hrm, I totally forgot about that site while we were discussing this
yesterday.  We only see three "issues" there, and that shows nobody
bothered to register issues, and it is very understandable.  People who
read the list know that all communication that is important to Git happen
here, and it will be an additional burden for reporters if they have to go
there (be it Jan's site, Andrew's one, or the issue system at GitHub for
that matter) and put what you already have posted.  People who do not read
the list had no chance knowing about Jan's site to begin with, given that
even I didn't immediately remember.

Also more importantly, the issues posted here are picked up and acted on
reasonably quickly---it often is more than "reasonably" quickly and I
often find myself looking at a new initial report, analysis of the problem
and a tested patch in a single thread, when I check my mailbox the morning.

Such an issue won't hit the tracker, and neither the initial reporter nor
the developers who responded should not do a lot of extra work to get the
thread in a "bug tracker" system.

Something based on the idea mentioned in Ævar's message (downstream in
this thread) to seamlessly integrate with the e-mail traffic might have a
chance to succeed.  I also think the integration must be two-way for it to
be useful.  A summary of "new issues untouched for N weeks" and another
"older issues unclosed for N weeks" periodically sent here, or something.

Perhaps collecting messages based on a handful of simple heuristics like
"A message mentioned the keyword 'bug', but no In-Reply-To for it from any
list regulars came in two weeks" might be a good place to start.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Neal Kreitzinger]

On 3/1/2012 11:10 AM, Junio C Hamano wrote:
>
> Something based on the idea mentioned in Ævar's message (downstream in
> this thread) to seamlessly integrate with the e-mail traffic might have a
> chance to succeed.  I also think the integration must be two-way for it to
> be useful.  A summary of "new issues untouched for N weeks" and another
> "older issues unclosed for N weeks" periodically sent here, or something.
>
> Perhaps collecting messages based on a handful of simple heuristics like
> "A message mentioned the keyword 'bug', but no In-Reply-To for it from any
> list regulars came in two weeks" might be a good place to start.
>
Why don't you just use git for your bug-tracking?  With commit-messages, 
annotated-tags, signed-tags, branches, branch-descriptions, git-notes, 
submodules, sha-1's, send-email, patches, hooks, etc., not to mention 
all-kinds-of-files you can commit (screen-shots, bug-reports, etc.) it 
seems you have more than enough pieces to build a bug-tracker with git. 
  For example:

Some Ideas:

(Setup)
   - The bug tracker repo could be a one-way mirror of git.git.  That 
way the bug-reports are literally tied to the source but don't clutter 
git.git.
   - Create a bug tracking branch parallel to each release branch.
   - Create a bug-branch off-of the buggy-commit and do the fix and make 
the bug-tracking-admin updates around the bug branch.

(Submitting Bugs)
   - Post-receive hook to assess new commits and send formatted email to 
mailing list, ie. new-bug push sends bug email to mailing list.
   - The mailing list creates "bug-report-commits" kind-of-like Junio 
mentioned above so the mailing list also feeding into it.
   - Bugs can be submitted directly to the bug repo and replicated to 
the mailing list via post-receive hook, and bugs can be submitted via 
the mailing list and automatically "forwarded" to the bug repo as "bug 
report" refs.
   - Screenshots could be a submodule for "large file" optimization.

(Issue-Numbers)
   - The sha-1 of the originating bug-report could be its issue-id.
   - Tag the commit with a "bug tag".  The sha-1 of the bug-tag could be 
the issue-id.
   - Calculate issue number from commit-counting to assign unique numbers.
   - The sha-1 of the buggy-commit could be the issue-id.

(Status Updates)
   - Use git-notes to update issue status.
   - Use annotated-tags to update issue status.
   - Use commit-messages to update issue status (or a file in the commit).

(Reporting)
   - Interrogate bug-formatted refnames with wildcards to generate 
queries of whats going on.
   - Use gitk for ad hoc or pre-defined gui queries of what-you-want-to-see.
   - Use git time-machine commands to see how old stuff is or whats new.
   - have a bug-report-id.txt file that people just modify and commit 
and use git-annotate to see the history of that bug-report.

(Signoff)
   - Signed tags for authenticity.

(Standardization)
   - Git-commands or git-aliases or git-contrib-scripts to make sure 
people do-it-right (format names/messages/files correctly), ie. templates.
   - Commit hook snippet that detects bug branch format and brings up 
bug commit template.
   - Rebase hook snippet that doesn't allow bug branches to be rebased.

(Integration)
   - The bug tracking system becomes part of the release workflow, ie., 
bugs are only considered fixed when they graduate to an official release 
or something close.
   - The "bug" tracking system also tracks "enhancements" and they are 
integrated into new releases using that workflow.

(Concurrency control)
   - Concurrent bug-report updates merge-conflict.
   - If someone beat-you-to-it then you cherry-pick or rebase on top of 
them.

(Interfaces)
   - works from commandline
   - works with gitk, git-gui
   - works with gitweb

(Quality)
   - Do you really need the greatest commercial-bug-tracker "x" or 
opensource-bug-tracker "y" to meet your needs?  If you've done well with 
nothing then you probably don't need all the bells an whistles. 
Management-by-report is mainly for clueless managers who want to feel 
like they know whats going on, and to supervise people who 
don't-really-care.  Since git people are on-the-ground and know whats 
going on, and are highly motivated individuals you probably don't really 
need all that fluff.  I suspect you would already be using 
opensource-bug-tracker "y" if there was one you actually liked using.
   - There's probably someone on this list who thinks they can create a 
bug-tracker with git that's better-than-anything-else-out-there, and 
they're probably right.

(Intuitive)
   - If you know how to use git then you know how to use git bug-tracker.
   - If you don't know how to use git then you probably shouldn't be 
submitting bug reports.

(Open Source Reusability)
   - The git bug repo system can then be used by others internally as 
they apply that system to their own project's canonical repository to 
track their bugs internally.

Some-Stuff-You-Already-Know:

(Distributed)
   - clone, push, pull -- your using git.

(Reliability)
   - you're using git so you know its uncorrupted, and if it does go 
down just clone from someone else and verify the checksum.

(Support)
   - git@vger.kernel.org
   - If git.git is down there is not much point in bug-tracking it.

I've been wanting to do this internally for a whle, but I'm still 
learning the git scm part.  Some of the git technologies I listed above 
I haven't used yet so they may sound a bit off, but I think you can get 
the gist of it.

I think git people in general would be enthusiastic about this, and if 
so, that would be the one thing that truly distinguishes this proposal 
from the others ;-)  You could call the repo or system bug-git, 
git-err-done, or something like that.

v/r,
neal

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Jonathan Nieder]

Neal Kreitzinger wrote:

> Why don't you just use git for your bug-tracking?
[...]
> I think git people in general would be enthusiastic about this

Not this git person. ;-)  I think there is a pretty major mismatch
between most VCSen's features and what a bugtracker needs.

That said, a good distributed bugtracker (which implies solving hard
social problems like "what to do if different contributors disagree on
severity" and simple technical problems like "how to present a
coherent conversation based on threads by people who might not have
been aware of each other") would be a very nice thing to see,
regardless of the choice of storage and network protocol used to back
it.

> git-err-done

Heh.

Thanks for some food for thought,
Jonathan

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Jonathan Nieder <jrnieder@gmail.com> writes:

> That said, a good distributed bugtracker (which implies solving hard
> social problems like "what to do if different contributors disagree on
> severity" and simple technical problems like "how to present a
> coherent conversation based on threads by people who might not have
> been aware of each other") would be a very nice thing to see,
> regardless of the choice of storage and network protocol used to back
> it.

Exactly. In the discussion of the "tracker" context, the choice of storage
medium is secondary.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Neal Kreitzinger]

On 3/1/2012 10:19 PM, Jonathan Nieder wrote:
>
> That said, a good distributed bugtracker (which implies solving hard
> social problems like "what to do if different contributors disagree on
> severity"
How do you solve that now?  You can have a field in the bug-report.txt 
for "severity" or a severity-file in the commit.  If people keep 
changing it and don't listen to the appointed tie-breaker then those 
people's patches probably aren't going to get applied anyway.

>   and simple technical problems like "how to present a
> coherent conversation based on threads by people who might not have
> been aware of each other") would be a very nice thing to see,
> regardless of the choice of storage and network protocol used to back
> it.
You can have a field in bug-report.txt called "discussion thread" or a 
file called discussion-thread.txt that contains the link to the mailing 
list for that thread (even terminal emulators automatically bring up 
links when you click on them).  People would have discussions there like 
they do now.  Real progress updates would be recorded in git.  
Discussions would remain on the mailing list.  Thread posts could also 
be recorded as git-notes.

Please let me know what else is a "hard social problem".  I'm not a 
bug-tracker expert but I've used a few bug-tracking systems and worked 
plenty of bugs though never on an opensource project so maybe I'm not 
aware of the "hard social problems".

That being said, I was in a position where I fixed on average over one 
bug per day (reported, fixed, tested, moved to production) for a total 
of 2,000+ bugfixes over about 6 years.  The most in one day was about 
eight (1 per hour).  It was a home-grown bugtracking system based on 
lotus-notes (ie, email integration).  Git has send-mail and the mailing 
list.  The workflow was:

(1) User submits bug-report (or enhancement-request).  Status = 
Unassigned,  Severity = whatever.
(2) Manager reviews bug-report and assigns to developer or rejects it.  
Status = Development or Rejected, Severity = whatever (manager might 
change it).
(3) Developer makes fix and marks ready for QA.  Status = QA.
(3.1)  during development Developer makes comments that are sent as 
emails and User reciprocates.
(4) User/tester tests fix and gives signoff.  Status = Move-to-Prod.
(4.1)  during testing the user/tester and developer make comments that 
are sent as emails.
(5) Developer promotes fix to production.  Status = Completed.
(6) Manager receives notice and acknowledges.  Status = Closed.

Here's how that might work with git:

(1) User (john.doe@nowhere.com) submits bug-report (or request).  Status 
= Unassigned,  Severity = whatever.
[BUG]  (starts the thread)

(2) Manager (git-maintainer) reviews bug-report and assigns to developer 
(or lieutenant) or rejects it.  Status = Development or Rejected, 
Severity = whatever.
[DEV] | [REJ]  (assigned to developer/lieutentant or rejected)

(3) Developer makes fix and marks ready for QA.  Status = QA.
(3.1)  during development Developer makes comments that are sent as 
emails and User reciprocates. (makes posts to thread on mailing list)
[QA]  (fix is in next)

(4) User/tester tests fix and gives signoff.  Status = Move-to-Prod.
(4.1)  during testing the user/tester and developer make comments that 
are sent as emails. (makes posts to thread on mailing list)
[PAS]  (fix passed testing)

(5) Developer promotes fix to production.  Status = Completed.
[CMP]  (fix is in rc)

(6) Manager (git-maintainer) receives notice.  Status = Closed.
[CLS]  (fix is in release)

I realize this is not an exact match of the git-workflow, but you get 
the idea.  I'm also new to mailinglists so I'm not sure if you can 
change part of the subject line.  If not, a header in the body could 
possibly be used.

Thank you for your time and consideration.

v/r,
neal

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Jonathan Nieder]

Neal Kreitzinger wrote:

> Please let me know what else is a "hard social problem".  I'm not a
> bug-tracker expert but I've used a few bug-tracking systems and
> worked plenty of bugs though never on an opensource project so maybe
> I'm not aware of the "hard social problems".

I was not alluding to bug ping-pong (which is a security or discipline
problem) but the task of reconciling independent good-faith updates
(which is a coordination problem).

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Neal Kreitzinger <nkreitzinger@gmail.com> writes:

> I realize this is not an exact match of the git-workflow, but you get
> the idea.  I'm also new to mailinglists so I'm not sure if you can
> change part of the subject line.  If not, a header in the body could
> possibly be used.

The most important information is missing from your discussion: who are
you trying to help, and what problem are you trying to solve?

When somebody posts a bug report to the list, with the current workflow,
one of these things happens:

 1. It is an already solved issue. People who are familiar with the
    existing fix may immediately answer, after running "git log", with "It
    is fixed in v1.7.6". Or somebody not so familiar with the fix may
    start "Does not reproduce for me who use the 'master' version. Git
    from what era are you using?" conversation. I do not think a bug
    tracker will help much in this case [*1*].

 2. It is an already answered non-issue. People who are familiar with the
    previous discussion may point at the list archive, or somebody may dig
    up the answer in the gmane archive. I do not know if a bug tracker
    will help much in this case. Having a place to point people at is
    better than having to write everything from scratch every time, but
    (1) looking for the previous discussion is the more time consuming
    part, and (2) once the previous discussion is found in the list
    archive, we already have the necessary pointer.

 3. People who are familiar with the area of the problem may start "Need
    more info" conversation. This may result in either finding the report
    a non-issue (#1 or #2), or it may turn out to be a real issue, and
    after further analysis, design and coding, may result in a fix.  Once
    this flow starts rolling, the current workflow works very well.

 4. It falls through cracks, because nobody even categorizes it into the
    above three.

I think the primary thing people want out of a bug tracker is to reduce
the frequency of #4.  The real solution for it is to free up time from
people who can do the later part of #3 so that they can spend more time to
turn #4 into #3.

A way to do so is for members of the community who are capable of doing #1
and #2 but not familiar enough with the code to do the later part of #3 to
help with earlier part of #3 (i.e. triaging).

As I already said. the mailing-list based workflow serves us reasonably
well once the ball is rolling in #3, and that was the reason why I
suggested some heuristics to catch #4 in my previous message.  There are
cases where the original reporter disappears during the "need more info"
exchange, and in such a case a tracking system _may_ be able to help us
remember that the issue is unresolved because of reporter inaction, but
the tracker won't respond to "need more info" itself, and people tend to
ignore automated nag mails, so there is still a need for warm body human
bug secretary who interfaces with the reporter in such a case.

In any case, any solution that demands more things to be done by people
near the core developers than they currently are already doing will make
things worse by exacerbating the problem that comes from a bottleneck in
the process.  I do not think your "The maintainer triages and assigns
issues to other developers" or "The assigned developer marks the issue as
'done' after fixing it" will fly very well, regardless of the use of any
bug tracker.


[Footnote]

 *1* If the symptom is so straightforward that a simple search in a bug
     tracker can produce hits for an already solved issue, grepping in
     Release Notes should equally work well.

 *2* I do not know if this happens too often to be a real problem, though.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andreas Ericsson]

On 03/02/2012 08:03 AM, Junio C Hamano wrote:

... a very concise and exact response.

> 
> In any case, any solution that demands more things to be done by people
> near the core developers than they currently are already doing will make
> things worse by exacerbating the problem that comes from a bottleneck in
> the process.  I do not think your "The maintainer triages and assigns
> issues to other developers" or "The assigned developer marks the issue as
> 'done' after fixing it" will fly very well, regardless of the use of any
> bug tracker.
> 

It works very well when there's the incentive of roof over one's head
and food on one's table to take care of the assigned issues. However,
nothing stops a git developer from saying "sorry, I'm busy" when being
assigned really, really boring tasks that they really don't feel like
doing.

One thing I could see a bugtracker would be good for is to get companies
that use git to vote on issues or features using real money. Developers
can then pick up the issue and do something with them.

Apart from that, I doubt there's much incentive for the people who do
any of the work to pick up issues nobody cares about. The number of bugs
falling through the cracks is too small to go through a lot of work just
to keep track of them, and the ones that do are ones that are primarily
of the bikeshedding variant or such weird corner-cases that they don't
happen in 99.999% of all use-cases git was designed for and is bid to
handle.

-- 
Andreas Ericsson                   andreas.ericsson@op5.se
OP5 AB                             www.op5.se
Tel: +46 8-230225                  Fax: +46 8-230231

Considering the successes of the wars on alcohol, poverty, drugs and
terror, I think we should give some serious thought to declaring war
on peace.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Junio C Hamano]

Andreas Ericsson <ae@op5.se> writes:

> On 03/02/2012 08:03 AM, Junio C Hamano wrote:
>
> ... a very concise and exact response.
>
>> In any case, any solution that demands more things to be done by people
>> near the core developers than they currently are already doing will make
>> things worse by exacerbating the problem that comes from a bottleneck in
>> the process.  I do not think your "The maintainer triages and assigns
>> issues to other developers" or "The assigned developer marks the issue as
>> 'done' after fixing it" will fly very well, regardless of the use of any
>> bug tracker.
>
> It works very well when there's the incentive of roof over one's head
> and food on one's table to take care of the assigned issues.

Your "this is a volunteer effort and assignment does not work like corp
environment" is valid, but I think it is missing the point.  A solution
that demands more from people who are already bottlenecks will not work
very well, even in a corporate environment where you have stronger
incentive to fill your assigned role.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Andrew Ardill]

Just a quick status update on where the issue tracker experiment at
git-scm.atlassian.net is up to:

The basic structure of the issue tracker is set up, and is ready for
people to log bugs if they want to. So far no one seems to inclined,
but note that it is still locked down to only people who have both
registered, and been promoted to 'trusted'. I will promote everyone
who has registered so far to 'trusted' in the hope that some of them
might add bugs!

The next main step is to have conversations on the list automatically
converted into issues that can be tracked. This will be either by
forwarding selected threads to the tracker, or forwarding everything
to the tracker and managing it from there. Unfortunately, a recent
upgrade broke the (unsupported) ability to create issues from
emails[1]. This will hopefully be fixed soon, and when it is we will
be able to move this experiment to the next phase.

If anyone has any ideas they would like to test, please let me know!

Regards,

Andrew Ardill

[1] Reply from Atlassian:
Officially we do not support
(http://confluence.atlassian.com/display/AOD/Restricted+Functions+in+Atlassian+OnDemand)
issue creation from email in OnDemand. The feature request for this
functionality is located at:
https://studio.atlassian.com/browse/JST-5649

There was the ability to create issues from email if emails were sent
to jira@<instance domain> as you tried to setup, but this
functionality was not officially supported and broke in the recent
JIRA 5 upgrade (which is why you receive those errors).

I believe we are planning to fix this in an upcoming bugfix release
within the next several weeks though, so please try again later.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Vincent van Ravesteijn]

Op 7-3-2012 9:03, Andrew Ardill schreef:
> Just a quick status update on where the issue tracker experiment at
> git-scm.atlassian.net is up to:
>
> The basic structure of the issue tracker is set up, and is ready for
> people to log bugs if they want to. So far no one seems to inclined,
> but note that it is still locked down to only people who have both
> registered, and been promoted to 'trusted'. I will promote everyone
> who has registered so far to 'trusted' in the hope that some of them
> might add bugs!

Done: https://git-scm.atlassian.net/browse/GIT-1

Vincent

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Joern Huxhorn]

On 02.03.2012, at 08:03, Junio C Hamano wrote:

> Neal Kreitzinger <nkreitzinger@gmail.com> writes:
> 
>> I realize this is not an exact match of the git-workflow, but you get
>> the idea.  I'm also new to mailinglists so I'm not sure if you can
>> change part of the subject line.  If not, a header in the body could
>> possibly be used.
> 
> The most important information is missing from your discussion: who are
> you trying to help, and what problem are you trying to solve?
> 
> When somebody posts a bug report to the list, with the current workflow,
> one of these things happens:
> 
> 1. It is an already solved issue. People who are familiar with the
>    existing fix may immediately answer, after running "git log", with "It
>    is fixed in v1.7.6". Or somebody not so familiar with the fix may
>    start "Does not reproduce for me who use the 'master' version. Git
>    from what era are you using?" conversation. I do not think a bug
>    tracker will help much in this case [*1*].
> 
> 2. It is an already answered non-issue. People who are familiar with the
>    previous discussion may point at the list archive, or somebody may dig
>    up the answer in the gmane archive. I do not know if a bug tracker
>    will help much in this case. Having a place to point people at is
>    better than having to write everything from scratch every time, but
>    (1) looking for the previous discussion is the more time consuming
>    part, and (2) once the previous discussion is found in the list
>    archive, we already have the necessary pointer.
> 
> 3. People who are familiar with the area of the problem may start "Need
>    more info" conversation. This may result in either finding the report
>    a non-issue (#1 or #2), or it may turn out to be a real issue, and
>    after further analysis, design and coding, may result in a fix.  Once
>    this flow starts rolling, the current workflow works very well.
> 
> 4. It falls through cracks, because nobody even categorizes it into the
>    above three.
> 
> I think the primary thing people want out of a bug tracker is to reduce
> the frequency of #4.  The real solution for it is to free up time from
> people who can do the later part of #3 so that they can spend more time to
> turn #4 into #3.
> 
> A way to do so is for members of the community who are capable of doing #1
> and #2 but not familiar enough with the code to do the later part of #3 to
> help with earlier part of #3 (i.e. triaging).
> 
> As I already said. the mailing-list based workflow serves us reasonably
> well once the ball is rolling in #3, and that was the reason why I
> suggested some heuristics to catch #4 in my previous message.  There are
> cases where the original reporter disappears during the "need more info"
> exchange, and in such a case a tracking system _may_ be able to help us
> remember that the issue is unresolved because of reporter inaction, but
> the tracker won't respond to "need more info" itself, and people tend to
> ignore automated nag mails, so there is still a need for warm body human
> bug secretary who interfaces with the reporter in such a case.
> 
> In any case, any solution that demands more things to be done by people
> near the core developers than they currently are already doing will make
> things worse by exacerbating the problem that comes from a bottleneck in
> the process.  I do not think your "The maintainer triages and assigns
> issues to other developers" or "The assigned developer marks the issue as
> 'done' after fixing it" will fly very well, regardless of the use of any
> bug tracker.
> 
> 
> [Footnote]
> 
> *1* If the symptom is so straightforward that a simple search in a bug
>     tracker can produce hits for an already solved issue, grepping in
>     Release Notes should equally work well.
> 
> *2* I do not know if this happens too often to be a real problem, though.

Sorry for the full quote.

I think the main problems with all issue trackers I know about are
- they are centralized, i.e. like SVN
- they all have a more or less clumsy web interface. All of them different, most of them configurable.

To get accepted in this community, an issue tracker would need to be decentralized (obviously including the ability to merge issue state and so on, likely git-based, probably simply included in the normal git repository of a project or in a separate issues-branch) and require a proper command line interface so it is properly scriptable (to feed it with threads from this mailing list, for example).

I'd love such a system.

I disagree with your assumption in 1. and *1*. A proper issue tracker has the ability to attach additional info (like stack traces or even a workaround) and files. Those would be searched, too, and would most likely not show up in the 'git log'.
An issue tracker could therefore remove noise from the mailing list which would automatically reduce workload on the core developers.

Your points 1. 2. and 3. all include "People who are familiar", i.e. it always involves some action from more or less cory developers. In case of an issue tracker, even duplicates and already solved issues (1. and 2.) serve a purpose since they'd indicate that either the previous solution hasn't been communicated good enough or that the duplicated issue has a high enough impact to creep up again.
Somebody needs to keep in mind all issues, in some way. I really wonder how you are able to cope with this task.

I'm not trying to lecture you that you are "doing it wrong". Your mailing-list based workflow seems to work incredibly well for you. But it doesn't really help to entice people into the community, either.

I'm (mostly) a silent observer of this list and just a (super-happy) user of git - but I really have a hard time getting a grip on the changes introduced in the different versions. You could obviously argue that I'm not trying hard enough since I'm not reading the 'git log' and all the diffs but I suspect that I'm already trying harder than most users.

For example, what happened to the git generation numbers discussed (in part) over here: http://comments.gmane.org/gmane.comp.version-control.git/177146
Are they already included in a released git version? If so, how would I find them? If not, why not? An issue tracker would be able to answer me, as a little-above-casual user, this question without resorting to asking you.

So I guess it all boils down to waiting until somebody is sufficiently annoyed by the current state of issue trackers and thus tries to implement a properly decentralized one that is likely based on plain text files (for easy merges) and features a proper way to query issues.

Just like Linus was sufficiently annoyed by the state of VCS.

Cheers,
Joern.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Jonathan Nieder]

Joern Huxhorn wrote:

> To get accepted in this community, an issue tracker would need to be
> decentralized

No, I don't think that's a requirement.

Thanks for your interest,
Jonathan

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Joern Huxhorn]

On 07.03.2012, at 14:53, Jonathan Nieder wrote:

> Joern Huxhorn wrote:
> 
>> To get accepted in this community, an issue tracker would need to be
>> decentralized
> 
> No, I don't think that's a requirement.
> 

I didn't mean to imply that this would be a matter of principle. A decentralized tracker would just offer the same advantages that a decentralized VCS has to offer. One could work on issues while not connected to the internet, for example. So this would be a VeryGoodThing™.

I don't think that it would be a good idea to "teach" git issue tracker functionality. I simply wanted to voice my opinion that something like the ideas Neal suggested down the thread, a cli issue tracker with git in the backend, are very worthwhile to evaluate. But it should have an additional layer so it's really easy to use. It has to be so it *is* actually used. It needs to reduce workload instead of increasing it.

Cheers,
Joern.

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Pau Garcia i Quiles]

On Wed, Mar 7, 2012 at 2:04 PM, Joern Huxhorn <jhuxhorn@googlemail.com> wrote:

> To get accepted in this community, an issue tracker would need to be decentralized (obviously including the ability to merge issue state and so on, likely git-based,
> probably simply included in the normal git repository of a project or in a separate issues-branch) and require a proper command line interface so it is properly
> scriptable (to feed it with threads from this mailing list, for example).
>
> I'd love such a system.

Take a look at Veracity ( http://veracity-scm.com/ )

Also, do not forget issue tracking must be possible for people who do
not use git. That's why we use git hosted at Assembla (
http://www.assembla.com ) at work: 2/3 of the people in the project
are not developers but marketing, verification, validation, support,
trainers, management, etc.

-- 
Pau Garcia i Quiles
http://www.elpauer.org
(Due to my workload, I may need 10 days to answer)

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Phil Hord]

On Fri, Mar 2, 2012 at 2:03 AM, Junio C Hamano <gitster@pobox.com> wrote:
> Neal Kreitzinger <nkreitzinger@gmail.com> writes:
>
>> I realize this is not an exact match of the git-workflow, but you get
>> the idea.  I'm also new to mailinglists so I'm not sure if you can
>> change part of the subject line.  If not, a header in the body could
>> possibly be used.
>
> The most important information is missing from your discussion: who are
> you trying to help, and what problem are you trying to solve?

Problems this could help solve (regardless of whether it's an
appropriate tool for the job):

1. Collects issues into a more concise list than the mailing list provides.

2. Collects issues (and discussion) conveniently bundled with the git
source code.

3. Collects issues for off-line reference and searching.

4. Reduction of list noise, if issues in git.git turn out to be better
grep-targets than the mailing list.

5. Serves as an incubator for a git-based distributed issues tracker
Best Practice or Dire Warning, depending on how it goes.

The current mailing list bug tracker, where finding existing issues
and previous discussions is "crowd-sourced" to the list, is very
efficient for the new users, but not so efficient for the core
developers and respondents.

I doubt this idea is really workable or appropriate for git.git, for
various reasons.  But I do think a well-designed, distributed,
git-based issue tracker could be useful for many other projects.  Many
others have tried and failed, so I am probably wrong about this last
statement.   See [*1*] for a list of mostly stagnating prior art.

Phil

[*1*] http://dist-bugs.branchable.com/software/

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Carlos Martín Nieto]

On Wed, 2012-02-29 at 12:19 -0500, opticyclic wrote:
> Firstly, why is there no Bug Tracker such as JIRA for the git project?

Bug trackers aren't magical. Just because a bug is in some sort of
database, it doesn't mean that it's going to get fixed faster. People
work on what they find interesting or their sponsors find important. Bug
databases are also notorious for getting filled with duplicate entries
and languishing bugs that are waiting for the original reporter to
answer with some information that the developers asked. Everyone would
also need to get an account on that bug tracker, making it harder to
report and contribute.


There used to be a wiki for buglets so people could get started, but I'm
not sure if it survived the k.org compromise.

> This mailing list is next to useless for users since searching is
> difficult, as is commenting and voting.

All you need to comment is an e-mail program, which most people have.
gmane also allows you to post from the web interface. What voting are
you referring to? There is no form of formal voting that isn't
restricted to the people responsible (or knowledgeable about) a
particular part of the project. And that's not even really voting, but a
review on the soundness of the patch.

> 
> Secondly, since one of the alleged reasons for creating git was to not
> have to deal with patches, why are pull requests disable and patches
> sent to this mailing list?!

Who said git was made to stop dealing with patches? Some of the git
terminology is influenced by that (compare 'git revert' with the idea of
a revert that other systems have). Do you follow the linux mailing list?
It's full of patches waiting to be reviewed.

git does use pull requests. That's how gitk and git-svn are updated, in
the git repository. I believe the git-subtree inclusion request also
took form of a pull request.

> I have read https://github.com/gitster/git/blob/master/Documentation/SubmittingPatches
> and it doesn't explain it.
> 
> I'm sure I don't have to tell you that GitHub has discussions on pull
> requests, which are easier to view than the mailing list archives.

Easier to view? Do you mean it's easier on the eyes? Easier to get an
overview? What do you want to view in these discussions? Why do you want
everyone to need a GitHub account to participate in git?

The GitHub web-UI has no threading, which means that you either discuss
all the patches together in one line so it's no longer clear who's
answering what, or you comment on the commit itself, which means that
the whole discussion is at least as segmented as what you get via
e-mail, and you have to scroll more to get to the discussion of a
particular commit.

TL;DR this is the way we've found to be the most effective.

   cmn

Re: Why Is There No Bug Tracker And Why Are Patches Sent Instead Of Pull Requests

[Sitaram Chamarty]

On Wed, Feb 29, 2012 at 10:49 PM, opticyclic <opticyclic@gmail.com> wrote:

> I'm sure I don't have to tell you that GitHub has discussions on pull
> requests, which are easier to view than the mailing list archives.

For some definition of "easier".  Personally, I loathe the interface.

You don't seem to realise that using that will force everyone to use
that same interface, rather than their choice of email clients.

I run a project that is mainly hosted on github, but I absolutely
positively refuse to use their web interface for anything.  Logging in
to check, instead of just reacting to email as usual, is a pain but
that is not all.

The issues system does have an email interface, but it is not a
substitute for email. I can't cc anyone else when I want to, for
instance (well I can, but any response the original requester then
makes using the website will not get cc-d to the person I cc-d, which
kinda defeats the whole purpose).

The pull system forces a --no-ff even if the merge is at the top of my
branch and doesn't need one. It also gives me no chance to fix up
minor typos, add any more text to the commit message, etc. (I can do
that afterward, but this forces a "push -f" or a trivial "typofix"
commit).

I want everything in *one* interface, and I want it just the way *I*
want it, and it shouldn't (necessarily) dictate how *you* should work.
 Email is just that.

sitaram