Sunseting mercurial

Sunseting mercurial

[Doug Goldstein]

Hello all,

Its been officially 5+ years since Xen has moved to git so I propose we
start thinking about when to retire the mercurial mirrors. At this point
the last stable version to be tracked in mercurial is 4.4 which is long
out of any form of support. I know some vendors still have support for
versions of Xen down to 4.1 but let's be realistic, there's not a flurry
of development happening in those old versions. The mercurial mirror is
often out of date (I know someone that's tried to use it) and in fact as
of this email its several weeks out of date.

So maybe its time we start thinking about sunsetting the mercurial
mirrors and use those resources for more practical uses.
-- 
Doug Goldstein

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Jan Beulich]

>>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
> Its been officially 5+ years since Xen has moved to git so I propose we
> start thinking about when to retire the mercurial mirrors. At this point
> the last stable version to be tracked in mercurial is 4.4 which is long
> out of any form of support. I know some vendors still have support for
> versions of Xen down to 4.1 but let's be realistic, there's not a flurry
> of development happening in those old versions. The mercurial mirror is
> often out of date (I know someone that's tried to use it) and in fact as
> of this email its several weeks out of date.
> 
> So maybe its time we start thinking about sunsetting the mercurial
> mirrors and use those resources for more practical uses.

This was brought up before, and I continue to agree _as long_ as
our web representation of the tree gains something similar to
hg's "annotate" functionality. Without that I find it quite hard to
locate commits most recently changing a line or an area of code.
Of course aiui this can be done from the command line, but only if
one happens to have a repo on the particular machine (which for
example I don't have or intend to have at home).

Speaking of mercurial, Anthony, wasn't it you who looks after the
git -> hg mirroring? The unstable.hg branch didn't get updated
after the last push from staging to master. Could you please see
if this can be rectified (assuming that my memory didn't fail me in
the first place, and it was someone else)?

Thanks, Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Roger Pau Monné]

On Mon, Mar 26, 2018 at 04:35:40AM -0600, Jan Beulich wrote:
> >>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
> > Its been officially 5+ years since Xen has moved to git so I propose we
> > start thinking about when to retire the mercurial mirrors. At this point
> > the last stable version to be tracked in mercurial is 4.4 which is long
> > out of any form of support. I know some vendors still have support for
> > versions of Xen down to 4.1 but let's be realistic, there's not a flurry
> > of development happening in those old versions. The mercurial mirror is
> > often out of date (I know someone that's tried to use it) and in fact as
> > of this email its several weeks out of date.
> > 
> > So maybe its time we start thinking about sunsetting the mercurial
> > mirrors and use those resources for more practical uses.
> 
> This was brought up before, and I continue to agree _as long_ as
> our web representation of the tree gains something similar to
> hg's "annotate" functionality. Without that I find it quite hard to
> locate commits most recently changing a line or an area of code.
> Of course aiui this can be done from the command line, but only if
> one happens to have a repo on the particular machine (which for
> example I don't have or intend to have at home).

AFAICT this could be enabled by setting:

$feature{'blame'}{'default'} = [1];

In the gitweb configuration file [0][1].

Roger.

[0] https://git-scm.com/docs/gitweb#gitweb-arguments
[1] https://git-scm.com/docs/gitweb#gitweb-blame

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Anthony PERARD]

On Mon, Mar 26, 2018 at 04:35:40AM -0600, Jan Beulich wrote:
> >>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
> > Its been officially 5+ years since Xen has moved to git so I propose we
> > start thinking about when to retire the mercurial mirrors. At this point
> > the last stable version to be tracked in mercurial is 4.4 which is long
> > out of any form of support. I know some vendors still have support for
> > versions of Xen down to 4.1 but let's be realistic, there's not a flurry
> > of development happening in those old versions. The mercurial mirror is
> > often out of date (I know someone that's tried to use it) and in fact as

Cardoe, if you mean "often out of date", that's true, but only by 10min
(which is the frequency at which a cron run).

> > of this email its several weeks out of date.

I'm guessing it's out of date just because I didn't thought of checking
if it was working after the recent server update. I only check every 2
months. I'm going to fix the script now.

> > So maybe its time we start thinking about sunsetting the mercurial
> > mirrors and use those resources for more practical uses.
> 
> This was brought up before, and I continue to agree _as long_ as
> our web representation of the tree gains something similar to
> hg's "annotate" functionality. Without that I find it quite hard to
> locate commits most recently changing a line or an area of code.
> Of course aiui this can be done from the command line, but only if
> one happens to have a repo on the particular machine (which for
> example I don't have or intend to have at home).
> 
> Speaking of mercurial, Anthony, wasn't it you who looks after the
> git -> hg mirroring? The unstable.hg branch didn't get updated
> after the last push from staging to master. Could you please see
> if this can be rectified (assuming that my memory didn't fail me in
> the first place, and it was someone else)?

Yes, that me that looks after it. I'll fix the mirror.

Thanks,

-- 
Anthony PERARD

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Anthony PERARD]

On Mon, Mar 26, 2018 at 11:56:49AM +0100, Anthony PERARD wrote:
> I'll fix the mirror.

It's fixed now.

-- 
Anthony PERARD

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Jan Beulich]

>>> On 26.03.18 at 15:18, <anthony.perard@citrix.com> wrote:
> On Mon, Mar 26, 2018 at 11:56:49AM +0100, Anthony PERARD wrote:
>> I'll fix the mirror.
> 
> It's fixed now.

Thanks a lot!

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Doug Goldstein]

On 3/26/18 5:35 AM, Jan Beulich wrote:
>>>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
>> Its been officially 5+ years since Xen has moved to git so I propose we
>> start thinking about when to retire the mercurial mirrors. At this point
>> the last stable version to be tracked in mercurial is 4.4 which is long
>> out of any form of support. I know some vendors still have support for
>> versions of Xen down to 4.1 but let's be realistic, there's not a flurry
>> of development happening in those old versions. The mercurial mirror is
>> often out of date (I know someone that's tried to use it) and in fact as
>> of this email its several weeks out of date.
>>
>> So maybe its time we start thinking about sunsetting the mercurial
>> mirrors and use those resources for more practical uses.
> 
> This was brought up before, and I continue to agree _as long_ as
> our web representation of the tree gains something similar to
> hg's "annotate" functionality. Without that I find it quite hard to
> locate commits most recently changing a line or an area of code.
> Of course aiui this can be done from the command line, but only if
> one happens to have a repo on the particular machine (which for
> example I don't have or intend to have at home).
> 

While I agree that gitweb should be changed to support showing that,
there exist a few options out there as well that work today.

https://gitlab.com/xen-project/xen/blame/staging/xen/common/Makefile
https://github.com/xen-project/xen/blame/staging/xen/common/Makefile

Browse to any file and click the Blame button or remember the URL and go
to it manually or here's a 10 second script.

#!/bin/bash -e

url="http://gitlab.com/xen-project/xen/blame"
thefile=$(readlink -f $1)
repo=$(git rev-parse --show-toplevel)
relfile=${thefile##${repo}}
branch=$(git rev-parse --abbrev-ref HEAD)
xdg-open "${url}/${branch}${relfile}"


-- 
Doug Goldstein

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[George Dunlap]

On Mon, Mar 26, 2018 at 2:56 PM, Doug Goldstein <cardoe@cardoe.com> wrote:
> On 3/26/18 5:35 AM, Jan Beulich wrote:
>>>>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
>>> Its been officially 5+ years since Xen has moved to git so I propose we
>>> start thinking about when to retire the mercurial mirrors. At this point
>>> the last stable version to be tracked in mercurial is 4.4 which is long
>>> out of any form of support. I know some vendors still have support for
>>> versions of Xen down to 4.1 but let's be realistic, there's not a flurry
>>> of development happening in those old versions. The mercurial mirror is
>>> often out of date (I know someone that's tried to use it) and in fact as
>>> of this email its several weeks out of date.
>>>
>>> So maybe its time we start thinking about sunsetting the mercurial
>>> mirrors and use those resources for more practical uses.
>>
>> This was brought up before, and I continue to agree _as long_ as
>> our web representation of the tree gains something similar to
>> hg's "annotate" functionality. Without that I find it quite hard to
>> locate commits most recently changing a line or an area of code.
>> Of course aiui this can be done from the command line, but only if
>> one happens to have a repo on the particular machine (which for
>> example I don't have or intend to have at home).
>>
>
> While I agree that gitweb should be changed to support showing that,
> there exist a few options out there as well that work today.
>
> https://gitlab.com/xen-project/xen/blame/staging/xen/common/Makefile
> https://github.com/xen-project/xen/blame/staging/xen/common/Makefile
>
> Browse to any file and click the Blame button or remember the URL and go
> to it manually or here's a 10 second script.

Jan, would the above two options be functional enough for you?

I agree that of all the things Anthony could be spending 10 minutes a
month on, maintaining a mercurial mirror isn't really a high priority.

 -George

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: Sunseting mercurial

[Jan Beulich]

>>> On 28.03.18 at 13:25, <dunlapg@umich.edu> wrote:
> On Mon, Mar 26, 2018 at 2:56 PM, Doug Goldstein <cardoe@cardoe.com> wrote:
>> On 3/26/18 5:35 AM, Jan Beulich wrote:
>>>>>> On 25.03.18 at 04:46, <cardoe@cardoe.com> wrote:
>>>> Its been officially 5+ years since Xen has moved to git so I propose we
>>>> start thinking about when to retire the mercurial mirrors. At this point
>>>> the last stable version to be tracked in mercurial is 4.4 which is long
>>>> out of any form of support. I know some vendors still have support for
>>>> versions of Xen down to 4.1 but let's be realistic, there's not a flurry
>>>> of development happening in those old versions. The mercurial mirror is
>>>> often out of date (I know someone that's tried to use it) and in fact as
>>>> of this email its several weeks out of date.
>>>>
>>>> So maybe its time we start thinking about sunsetting the mercurial
>>>> mirrors and use those resources for more practical uses.
>>>
>>> This was brought up before, and I continue to agree _as long_ as
>>> our web representation of the tree gains something similar to
>>> hg's "annotate" functionality. Without that I find it quite hard to
>>> locate commits most recently changing a line or an area of code.
>>> Of course aiui this can be done from the command line, but only if
>>> one happens to have a repo on the particular machine (which for
>>> example I don't have or intend to have at home).
>>>
>>
>> While I agree that gitweb should be changed to support showing that,
>> there exist a few options out there as well that work today.
>>
>> https://gitlab.com/xen-project/xen/blame/staging/xen/common/Makefile 
>> https://github.com/xen-project/xen/blame/staging/xen/common/Makefile 
>>
>> Browse to any file and click the Blame button or remember the URL and go
>> to it manually or here's a 10 second script.
> 
> Jan, would the above two options be functional enough for you?

In the worst case the above would probably work, but Ian has
turned on "blame" for xen.git already, which I much prefer over
using secondary repos.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel