[Xenomai-core] Bug tracker.

[Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Hi,

GNA offers a bug tracking system which is undoubtely a useful tool
for lots of projects. What about using it for the Xenomai project ?

The bug submission form is here:
https://gna.org/bugs/?func=additem&group=xenomai

I just had a look at its configuration, here is how it could be
configured if we agreed on using it :
- add a little explanation in place of the "0", reminding people to use
  mailing list/archives before posting a bug and what details are needed
  for a bug report (exact release or revision, kernel
  configuration file, xenomai configuration, small test case, kernel
  crash dump when appliable, etc...)
- add the fields which seem to be missing ; so far, only the software
  revision and component (nucleus, native skin, posix skin, etc...).

Suggestions and comments welcome.

-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Heikki Lindholm]

Gilles Chanteperdrix kirjoitti:
> Hi,
> 
> GNA offers a bug tracking system which is undoubtely a useful tool
> for lots of projects. What about using it for the Xenomai project ?

I'm not stronly opposing, but in my opinion it causes information to 
scatter in contrast to the ml. There'll be bug reports on the ml 
regardless of having or not having a bug tracking system in function. 
It's more difficult to follow/search two places than one. And this 
project isn't the size of openoffice or debian, so maybe the ml doesn't 
get cluttered by incoming bug reports. And finally, many bug reports 
already contain a reasonable fix and their tracking summarizes to 
"Applied." There are benefits, too, of course, like forcing a good 
format for a report.

-- Heikki Lindholm

Re: [Xenomai-core] Bug tracker.

[Romain Lenglet]

> I'm not stronly opposing, but in my opinion it causes
> information to scatter in contrast to the ml. There'll be bug
> reports on the ml regardless of having or not having a bug
> tracking system in function. It's more difficult to
> follow/search two places than one. And this project isn't the
> size of openoffice or debian, so maybe the ml doesn't get
> cluttered by incoming bug reports. And finally, many bug
> reports already contain a reasonable fix and their tracking
> summarizes to "Applied." There are benefits, too, of course,
> like forcing a good format for a report.

One point in favor of the use of a tracker like that of Gforge or 
GNA, is that it assigns a unique number to every bug / wish, 
which could then be easily referenced in Changelogs, CVS/SVN 
commits, etc.

It is easier to track bugs and have an history (when that bug has 
been reported? when has it been fixed? etc. which is not always 
clear from a ML).

I see it as complementary to the ML.


(my 2-yen contribution)

-- 
Romain Lenglet

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > Gilles Chanteperdrix kirjoitti:
 > > Hi,
 > > 
 > > GNA offers a bug tracking system which is undoubtely a useful tool
 > > for lots of projects. What about using it for the Xenomai project ?
 > 
 > I'm not stronly opposing, but in my opinion it causes information to 
 > scatter in contrast to the ml. There'll be bug reports on the ml 
 > regardless of having or not having a bug tracking system in
 > function. 

I do not see it as a substitute for the mailing list. Of course there
would still be posts on the mailing list, but it provides a standardized
answer to bug reports on the mailing lists; instead of answering "please
send me your .xeno_config, your .config, a small example exhibiting the
behaviour that you think is bad, please look at the mailing list
archives, do you have the bug with the latest revision  ?" over and over
again, we would answer "please read instructions at http://.... and fill
a bug report". The bug data get naturally attached to it in the
tracker. Another reason to track bugs is simply, well, to avoid
forgetting them.


 > It's more difficult to follow/search two places than one. 

If for every real bug reported on the mailing list, there is a entry on
the tracker ; there is only one place to search : the tracker, because
it is a database, and much easier to search than mailing list
archives, provided of course we add the important fields to the
submission form.


 > And this project isn't the size of openoffice or debian, so maybe the
 > ml doesn't get cluttered by incoming bug reports. 

Bug reports and FAQs do represent the vast majority of RTAI user mailing
list traffic for example. Answering these is time consuming, so since
our ressources are limited, any "productivity tool" is a good idea. If
the bug tracker turns out to be time consuming, we will stop using
it, but how do we know if we do not try ?


 > And finally, many bug reports already contain a reasonable fix and
 > their tracking summarizes to "Applied." There are benefits, too, of
 > course, like forcing a good format for a report.

My perception is that bug reports on the mailing list almost never
contain a fix, and even worse, people naturally tend to avoid giving you
all the elements that would allow to solve the bug, so you have to ask
the same questions over and over again. 


-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Heikki Lindholm]

Gilles Chanteperdrix kirjoitti:
> Heikki Lindholm wrote:
>  > Gilles Chanteperdrix kirjoitti:
>  > > Hi,
>  > > 
>  > > GNA offers a bug tracking system which is undoubtely a useful tool
>  > > for lots of projects. What about using it for the Xenomai project ?
>  > 
>  > It's more difficult to follow/search two places than one. 
> 
> If for every real bug reported on the mailing list, there is a entry on
> the tracker ; there is only one place to search : the tracker, because
> it is a database, and much easier to search than mailing list
> archives, provided of course we add the important fields to the
> submission form.

There's the "if". What I've seen on sourceforge is that often times bugs 
that are reported on the ml don't appear in the tracker and vice versa, 
although the tracker can probably be configured to forward reports to 
ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
only allow developers to file bugs into the tracker?

>  > And this project isn't the size of openoffice or debian, so maybe the
>  > ml doesn't get cluttered by incoming bug reports. 
> 
> Bug reports and FAQs do represent the vast majority of RTAI user mailing
> list traffic for example. Answering these is time consuming, so since
> our ressources are limited, any "productivity tool" is a good idea. If
> the bug tracker turns out to be time consuming, we will stop using
> it, but how do we know if we do not try ?

As I said I'm not strongly opposing. Go ahead and give it a spin - not 
too difficult to back out from that. Although, on the USER side, I've 
never preferred a bts (login req/learn to use/etc) to ml (e-mail client).

> all the elements that would allow to solve the bug, so you have to ask
> the same questions over and over again. 

Isn't this what most of life is about anyway? :)

-- Heikki Lindholm

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > There's the "if". What I've seen on sourceforge is that often times bugs 
 > that are reported on the ml don't appear in the tracker and vice versa, 
 > although the tracker can probably be configured to forward reports to 
 > ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
 > only allow developers to file bugs into the tracker?

I would use it as a one line answer to bug reports on the mailing
list... I would prefer users to always post on the mailing list first,
and if it looks like a real bug, a developer would ask the user to fill
a bug report.

-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Heikki Lindholm]

Gilles Chanteperdrix kirjoitti:
> Heikki Lindholm wrote:
>  > There's the "if". What I've seen on sourceforge is that often times bugs 
>  > that are reported on the ml don't appear in the tracker and vice versa, 
>  > although the tracker can probably be configured to forward reports to 
>  > ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
>  > only allow developers to file bugs into the tracker?
> 
> I would use it as a one line answer to bug reports on the mailing
> list... I would prefer users to always post on the mailing list first,
> and if it looks like a real bug, a developer would ask the user to fill
> a bug report.

That might work. As a case study of something to avoid: I know at least 
three persistent bugs in the firefox browser I'm using and I thought 
about reporting them, ending up in 
http://www.mozilla.org/support/firefox/bugs
which in turn leads to many worksome steps and eventually to
http://www.mozilla.org/quality/bug-writing-guidelines.html
Blech! It seems easier to write a firefox from ground up than giving 
them a bug report --- and thus my observations stay unreported. Moral of 
the story: keep it simple!

-- Heikki Lindholm

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > Gilles Chanteperdrix kirjoitti:
 > > Heikki Lindholm wrote:
 > >  > There's the "if". What I've seen on sourceforge is that often times bugs 
 > >  > that are reported on the ml don't appear in the tracker and vice versa, 
 > >  > although the tracker can probably be configured to forward reports to 
 > >  > ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
 > >  > only allow developers to file bugs into the tracker?
 > > 
 > > I would use it as a one line answer to bug reports on the mailing
 > > list... I would prefer users to always post on the mailing list first,
 > > and if it looks like a real bug, a developer would ask the user to fill
 > > a bug report.
 > 
 > That might work. As a case study of something to avoid: I know at least 

Ok. Have a look and tell me:

https://gna.org/bugs/?func=additem&group=xenomai

-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Heikki Lindholm]

Gilles Chanteperdrix kirjoitti:
> Heikki Lindholm wrote:
>  > Gilles Chanteperdrix kirjoitti:
>  > > Heikki Lindholm wrote:
>  > >  > There's the "if". What I've seen on sourceforge is that often times bugs 
>  > >  > that are reported on the ml don't appear in the tracker and vice versa, 
>  > >  > although the tracker can probably be configured to forward reports to 
>  > >  > ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
>  > >  > only allow developers to file bugs into the tracker?
>  > > 
>  > > I would use it as a one line answer to bug reports on the mailing
>  > > list... I would prefer users to always post on the mailing list first,
>  > > and if it looks like a real bug, a developer would ask the user to fill
>  > > a bug report.
>  > 
>  > That might work. As a case study of something to avoid: I know at least 
> 
> Ok. Have a look and tell me:
> 
> https://gna.org/bugs/?func=additem&group=xenomai

Looks good, doesn't place too much burden on the user. One might 
consider adding "details of the hardware platform used 
(CPU/chipset/etc)" or similar, because that isn't necessarily obvious 
from the configs.

-- Heikki Lindholm

Re: [Xenomai-core] Bug tracker.

[Heikki Lindholm]

Gilles Chanteperdrix kirjoitti:
> Heikki Lindholm wrote:
>  > Gilles Chanteperdrix kirjoitti:
>  > > Heikki Lindholm wrote:
>  > >  > There's the "if". What I've seen on sourceforge is that often times bugs 
>  > >  > that are reported on the ml don't appear in the tracker and vice versa, 
>  > >  > although the tracker can probably be configured to forward reports to 
>  > >  > ml, or? Who will type the ml-only reports to the tracker? Btw. would you 
>  > >  > only allow developers to file bugs into the tracker?
>  > > 
>  > > I would use it as a one line answer to bug reports on the mailing
>  > > list... I would prefer users to always post on the mailing list first,
>  > > and if it looks like a real bug, a developer would ask the user to fill
>  > > a bug report.
>  > 
>  > That might work. As a case study of something to avoid: I know at least 
> 
> Ok. Have a look and tell me:
> 
> https://gna.org/bugs/?func=additem&group=xenomai

Btw. Do you think the Task/Patch/Tech Support Managers are worth using? 
If not, can the unused(/-ful) ones be disabled from the site?

-- Heikki Lindholm

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > Looks good, doesn't place too much burden on the user. One might 
 > consider adding "details of the hardware platform used 
 > (CPU/chipset/etc)" or similar, because that isn't necessarily obvious 
 > from the configs.

Done, thanks.

-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > Btw. Do you think the Task/Patch/Tech Support Managers are worth using? 

The patch manager may be useful, it avoids loosing patches. For the
rest, I have no opinion, maybe someone else?

-- 


					    Gilles Chanteperdrix.

Re: [Xenomai-core] Bug tracker.

[Gilles Chanteperdrix]

Heikki Lindholm wrote:
 > Btw. Do you think the Task/Patch/Tech Support Managers are worth using? 
 > If not, can the unused(/-ful) ones be disabled from the site?

We will wait for the web site to sort this out.

-- 


					    Gilles Chanteperdrix.