FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian Development

 
 
LinkBack Thread Tools
 
Old 02-17-2008, 08:02 PM
Scott Kitterman
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

On Sunday 17 February 2008 15:41, Raphael Geissert wrote:
> Hello all,
>
> [Please respect the Reply-To header]
>
> In order to bring some more QA on the watch files subject I'd like to start
> a permanent MBF on packages whose Debian upstream version (the version
> string from Version: without the epoch and the Debian revision) is higher
> than the latest upstream version found thanks to their watch file.
>
> Rationale: the watch files are meant to keep track of upstream and if
> there's a newer version not being reported by the watch file it means that
> it needs to be fixed.
>
> Please note that this situation often occurs when the maintainer didn't
> make the watch file strip some +VCSrevNNNNN that was added to the Debian
> Version.
>
> If nobody objects I'll start filling (in an automated way since there are
> no false positives) reports on the 307 source packages which report a
> Debian upstream version higher than Upstream version by the watch file.

I disagree. You list my package:

Scott Kitterman <scott@kitterman.com>
* *pysubnettree

The reason this package is in the state it's in is that upstream uploaded a
new version of the package without changing the released version number so a
fake version number was needed. While suboptimal, there is no bug in the
watch file.

While this is no doubt a rare condition, I believe that your assertion that
there are no false positives is incorrect.

Scott K
 
Old 02-17-2008, 08:57 PM
"Wesley J. Landaker"
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

On Sunday 17 February 2008 13:41:34 Raphael Geissert wrote:
> If nobody objects I'll start filling (in an automated way since there are
> no false positives) reports on the 307 source packages which report a
> Debian upstream version higher than Upstream version by the watch file.

I don't know what you mean by "there are no false positives". My ghdl
package you mention is a false positive, for one.

--
Wesley J. Landaker <wjl@icecavern.net> <xmpp:wjl@icecavern.net>
OpenPGP FP: 4135 2A3B 4726 ACC5 9094 0097 F0A9 8A4C 4CD6 E3D2
 
Old 02-17-2008, 09:09 PM
Raphael Geissert
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Hello,

Raphael Hertzog wrote:
> Hello,
>
> On Sun, 17 Feb 2008, Raphael Geissert wrote:
>> Rationale: the watch files are meant to keep track of upstream and if
>> there's a newer version not being reported by the watch file it means
>> that it needs to be fixed.
>>
>> Please note that this situation often occurs when the maintainer didn't
>> make the watch file strip some +VCSrevNNNNN that was added to the Debian
>> Version.
>>
>> If nobody objects I'll start filling (in an automated way since there are
>> no false positives) reports on the 307 source packages which report a
>> Debian upstream version higher than Upstream version by the watch file.
>
> I do object. I don't think it's really important to complicate watch files
> to strip .dfsg or +svnXXXX that are addded by Debian maintainers. The most
> important thing with watch files is that a new upstream version is
> detected... but it's not important if the report says that Debian is newer
> than upstream when in fact we're at the same version.

Ack, what about only reporting (thus in a non automated way) on those which
are not affected by any repackaging/similar version part?

Some examples:
package|Debian Version|Reported upstream version|Debian upstream version
xrn|9.02-7.1|1|9.02
swfdec-gnome|2.21.90-2|0.5.5|2.21.90
conduit|0.3.6-2|0.3.4|0.3.6
diction|uupdate|1.11|uupdate
eject|2.1.5-6|2.1.0|2.1.5
epiphany|0.7.0-1|0.6.1|0.7.0
at-spi|1.21.5-1|1.20.1|1.21.5
glib2.0|2.15.5-1|2.14.6|2.15.5
...and so on

>
> And when we have +svnXXXX we are indeed newer than the upstream released
> tarball and the information is correct! So stripping that part would be a
> mistake.

IMHO it would be better to strip that part with a dversionmangle. However,
DEHS currently compares with $upstream le $debian so those packages are
marked as up to date.

>
> Cheers,

Cheers,
Raphael


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-17-2008, 09:17 PM
Raphael Geissert
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Wesley J. Landaker wrote:

> On Sunday 17 February 2008 13:41:34 Raphael Geissert wrote:
>> If nobody objects I'll start filling (in an automated way since there are
>> no false positives) reports on the 307 source packages which report a
>> Debian upstream version higher than Upstream version by the watch file.
>
> I don't know what you mean by "there are no false positives". My ghdl
> package you mention is a false positive, for one.
>

Please refer to my note:
> Please note that this situation often occurs when the maintainer didn't
make
> the watch file strip some +VCSrevNNNNN that was added to the Debian
Version.

package|Debian Version |upstream version|Debian upstream version
ghdl|0.26+gcc4.1.2~dfsg-2|0.26|0.26+gcc4.1.2

Comparing the upstream version and the Debian upstream version will,
correctly, say that the latter is higher (in other words: it isn't a false
positive, the Debian upstream version _IS_ higher than upstream).
Please also note that as explained on[1] DEHS is additionally
stripping 'dfsg', 'debian', and 'ds' from the Debian Version, which is
something IMHO must be done (and I'm of the opinion that DEHS should not do
that but let uscan and the watch file do it, hence my RFC of a version=4
watch file format I'll soon be sending to -devel).


[1]http://wiki.debian.org/DEHS (section 'Debian repackaging (a.k.a. trust
what uscan says)')


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-17-2008, 09:22 PM
Raphael Geissert
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Scott Kitterman wrote:

> On Sunday 17 February 2008 15:41, Raphael Geissert wrote:
>>
>> Please note that this situation often occurs when the maintainer didn't
>> make the watch file strip some +VCSrevNNNNN that was added to the Debian
>> Version.
>>
>> If nobody objects I'll start filling (in an automated way since there are
>> no false positives) reports on the 307 source packages which report a
>> Debian upstream version higher than Upstream version by the watch file.
>
> I disagree. You list my package:
>
> Scott Kitterman <scott@kitterman.com>
> pysubnettree
>
> The reason this package is in the state it's in is that upstream uploaded
> a new version of the package without changing the released version number
> so a
> fake version number was needed. While suboptimal, there is no bug in the
> watch file.

If Debian's 0.11+1-1 is upstream's 0.11 why not just strip the '+1' using
dversionmangle?
That's in my POV the bug.

>
> While this is no doubt a rare condition, I believe that your assertion
> that there are no false positives is incorrect.
>
> Scott K

Cheers,
Raphael


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-17-2008, 09:35 PM
Scott Kitterman
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

On Sunday 17 February 2008 17:22, Raphael Geissert wrote:
> Scott Kitterman wrote:
> > On Sunday 17 February 2008 15:41, Raphael Geissert wrote:
> >> Please note that this situation often occurs when the maintainer didn't
> >> make the watch file strip some +VCSrevNNNNN that was added to the Debian
> >> Version.
> >>
> >> If nobody objects I'll start filling (in an automated way since there
> >> are no false positives) reports on the 307 source packages which report
> >> a Debian upstream version higher than Upstream version by the watch
> >> file.
> >
> > I disagree. You list my package:
> >
> > Scott Kitterman <scott@kitterman.com>
> > pysubnettree
> >
> > The reason this package is in the state it's in is that upstream uploaded
> > a new version of the package without changing the released version number
> > so a
> > fake version number was needed. While suboptimal, there is no bug in the
> > watch file.
>
> If Debian's 0.11+1-1 is upstream's 0.11 why not just strip the '+1' using
> dversionmangle?
> That's in my POV the bug.
>
I think rewriting watch files for one time events is a mistake. If this were
a permanent feature of the version numbering I would agree. I suppose the
easiest solution for me to not be bothered about this would be to remove the
watch file on the next upload.

Scott K


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-17-2008, 11:55 PM
Raphael Geissert
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Scott Kitterman wrote:
> On Sunday 17 February 2008 17:22, Raphael Geissert wrote:
>>
>> If Debian's 0.11+1-1 is upstream's 0.11 why not just strip the '+1' using
>> dversionmangle?
>> That's in my POV the bug.
>>
> I think rewriting watch files for one time events is a mistake. If this
> were
> a permanent feature of the version numbering I would agree.

The thing is, when you make such kind of uploads all you have to make sure
is that uscan still says your package is up to date.

> I suppose the
> easiest solution for me to not be bothered about this would be to remove
> the watch file on the next upload.

You won't be bothered if you also maintain the watch file.
And as I said in my response to Raphael Hertzog I could skip those where the
Debian version has something like +svn, +cvs, -pre, and also probably skip
those such as yours: +n.
But those I really don't want to exclude are the ones
having 'dsfg', 'ds', 'debian', or ones whose watch file really reports an
older version (e.g. in Debian: 2.3.1, upstream: 2.0.1).

>
> Scott K

Cheers,
Raphael


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-18-2008, 12:13 AM
Russ Allbery
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Raphael Geissert <atomo64@gmail.com> writes:

> In order to bring some more QA on the watch files subject I'd like to
> start a permanent MBF on packages whose Debian upstream version (the
> version string from Version: without the epoch and the Debian revision)
> is higher than the latest upstream version found thanks to their watch
> file.

> Russ Allbery <rra@debian.org>
> gnubg

Upstream stopped doing real releases a while back although hopefully will
again do some someday. Currently, all that's available is nightly
snapshots. I can:

* Keep pointing the watch file at the actual official release location in
the hope that upstream will eventually release a newer official version.

* Point the watch file at the daily snapshots and have it always be out of
date because I'm not going to release new versions of gnubg daily.

* Delete the watch file.

Which would you rather I do? I personally think the first option is the
best, which is why I'm doing it, but I don't care that much.

--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-18-2008, 12:24 AM
Scott Kitterman
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

On Sun, 17 Feb 2008 18:55:26 -0600 Raphael Geissert
<atomo64+debian@gmail.com> wrote:
>Scott Kitterman wrote:
>> On Sunday 17 February 2008 17:22, Raphael Geissert wrote:
>>>
>>> If Debian's 0.11+1-1 is upstream's 0.11 why not just strip the '+1'
using
>>> dversionmangle?
>>> That's in my POV the bug.
>>>
>> I think rewriting watch files for one time events is a mistake. If this
>> were
>> a permanent feature of the version numbering I would agree.
>
>The thing is, when you make such kind of uploads all you have to make sure
>is that uscan still says your package is up to date.
>
>> I suppose the
>> easiest solution for me to not be bothered about this would be to remove
>> the watch file on the next upload.
>
>You won't be bothered if you also maintain the watch file.
>And as I said in my response to Raphael Hertzog I could skip those where
the
>Debian version has something like +svn, +cvs, -pre, and also probably skip
>those such as yours: +n.

Fair enough.

>But those I really don't want to exclude are the ones
>having 'dsfg', 'ds', 'debian', or ones whose watch file really reports an
>older version (e.g. in Debian: 2.3.1, upstream: 2.0.1).
>

There are also packages where an upstream release is missing entirely.

This sounds more reasonable to me, but I think you should publish a revised
list and give maintainers a chance to respond.

Scott K


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 02-18-2008, 01:01 AM
Raphael Geissert
 
Default Proposed MBF: Debian upstream version higher than watch file-reported upstream version

Russ Allbery wrote:

> Raphael Geissert <atomo64@gmail.com> writes:
>
>> Russ Allbery <rra@debian.org>
>> gnubg
>
> Upstream stopped doing real releases a while back although hopefully will
> again do some someday. Currently, all that's available is nightly
> snapshots. I can:
>
> * Keep pointing the watch file at the actual official release location in
> the hope that upstream will eventually release a newer official version.
>
> * Point the watch file at the daily snapshots and have it always be out of
> date because I'm not going to release new versions of gnubg daily.
>
> * Delete the watch file.
>
> Which would you rather I do? I personally think the first option is the
> best, which is why I'm doing it, but I don't care that much.
>

I would probably have kept the version number at 0.14.3 and append
+snapshotYYYYMMDD. However if I only were able to choose between those
three I would choose the first one.

But I do wonder why upstream hasn't released any version, and guessing
this:, but updated the changelog or a similar file you use to know that the
latest version is 0.16

By the way, I've been working on uscan and my changes can be expected to be
available within some days. One of the features I've been working on is
#395439 which might be helpful in your situation.

Cheers,
Raphael Geissert


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 

Thread Tools




All times are GMT. The time now is 07:02 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org