hg.python.org Mercurial upgrade
Hello, I've upgraded the Mercurial version on hg.python.org. If there any problems, don't hesitate to post here. (apart from the connectivity problems we seem to have from time to time and which shouldn't be related) Regards Antoine.
2013/1/22 Antoine Pitrou <solipsis@pitrou.net>
I've upgraded the Mercurial version on hg.python.org. If there any problems, don't hesitate to post here.
I've noticed a display glitch with the hg viewer: http://hg.python.org/cpython/rev/6df0b4ed8617#l2.8 There is a "[#14591]" link which causes the rest of the line to be shifted. -- Amaury Forgeot d'Arc
2013/1/23 Antoine Pitrou <solipsis@pitrou.net>
Also I noticed that the table at the top (author, date, etc) is not nicely aligned; the first column should be wider and "change baseline" (btw, what does this mean?) should not wrap. Experiments in Chrome suggest to change a line in static/style-paper.css, for the "#changesetEntry th" rule: "width: 8em;" -- Amaury Forgeot d'Arc
On Wed, 23 Jan 2013 21:36:21 +0100 "Amaury Forgeot d'Arc" <amauryfa@gmail.com> wrote:
I don't know, that's a new thing in hgweb apparently. The issue should be reported in http://bz.selenic.com/, since that's nothing python.org-specific AFAIR. Regards Antoine.
On Wed, Jan 23, 2013 at 9:43 PM, Antoine Pitrou <solipsis@pitrou.net> wrote:
FWIW this was an attempt to fix the links to issues in http://hg.python.org/cpython/. AFAIU the interhg extension used here to turn "#12345" to links affects at least 3 places: 1) the description of each changeset in the "shortlog" page (e.g. http://hg.python.org/cpython/); 2) the description at the top of the "rev" page (e.g. http://hg.python.org/cpython/rev/6df0b4ed8617); 3) the code in the "diff"/"rev"/"annotate" and possibly other pages (e.g. http://hg.python.org/cpython/rev/6df0b4ed8617#l2.6); With the previous solution, case 1 was broken, but links for cases 2-3 worked fine. The problem is that in 1 the description is already a link, so the result ended up being something like '<a href="rev/...">Issue <a href="b.p.o/12345">#12345</a> is now fixed</a>'. With the new solution 1-2 work (the links are added/moved at the end), but it's glitched for case 3. Unless interhg provides a way to limit the replacement only to specific places and/or use different replacements for different places, we will either have to live with these glitches or come up with a proper fix done at the right level. Best Regards, Ezio Melotti
Regards
Antoine.
On Wed, Jan 23, 2013 at 6:16 PM, Ezio Melotti <ezio.melotti@gmail.com> wrote:
How does the above relate to this issue? http://bugs.python.org/issue15919 --Chris
Antoine Pitrou <solipsis@pitrou.net> writes:
I'm not sure if this is related to the upgrade specifically (or just some other hg issue0, but yesterday my Windows 7 buildbot failed an hg pull operation on the 3.x branch and automatically removed it's local repository. Ever since then an attempt to start fresh (with an hg clone) is dying during the clone operation. I'm running the clone operation manually now, and after sitting there for 10 minutes or so at the "adding changesets" stage, it died just like in the buildbot logs ("transaction abort!" followed by "rollback completed"). Nothing was happening on the buildbot during the adding changesets delay so I'm assuming that was all server-side. -- David
Le Wed, 23 Jan 2013 17:37:53 -0500, David Bolen <db3l.net@gmail.com> a écrit :
Given the delay (10 minutes), I'm assuming this has more to do with the connectivity problem Noah is trying to investigate: http://mail.python.org/pipermail/catalog-sig/2013-January/004736.html Regards Antoine.
2013/1/22 Antoine Pitrou <solipsis@pitrou.net>
I've upgraded the Mercurial version on hg.python.org. If there any problems, don't hesitate to post here.
I've noticed a display glitch with the hg viewer: http://hg.python.org/cpython/rev/6df0b4ed8617#l2.8 There is a "[#14591]" link which causes the rest of the line to be shifted. -- Amaury Forgeot d'Arc
2013/1/23 Antoine Pitrou <solipsis@pitrou.net>
Also I noticed that the table at the top (author, date, etc) is not nicely aligned; the first column should be wider and "change baseline" (btw, what does this mean?) should not wrap. Experiments in Chrome suggest to change a line in static/style-paper.css, for the "#changesetEntry th" rule: "width: 8em;" -- Amaury Forgeot d'Arc
On Wed, 23 Jan 2013 21:36:21 +0100 "Amaury Forgeot d'Arc" <amauryfa@gmail.com> wrote:
I don't know, that's a new thing in hgweb apparently. The issue should be reported in http://bz.selenic.com/, since that's nothing python.org-specific AFAIR. Regards Antoine.
On Wed, Jan 23, 2013 at 9:43 PM, Antoine Pitrou <solipsis@pitrou.net> wrote:
FWIW this was an attempt to fix the links to issues in http://hg.python.org/cpython/. AFAIU the interhg extension used here to turn "#12345" to links affects at least 3 places: 1) the description of each changeset in the "shortlog" page (e.g. http://hg.python.org/cpython/); 2) the description at the top of the "rev" page (e.g. http://hg.python.org/cpython/rev/6df0b4ed8617); 3) the code in the "diff"/"rev"/"annotate" and possibly other pages (e.g. http://hg.python.org/cpython/rev/6df0b4ed8617#l2.6); With the previous solution, case 1 was broken, but links for cases 2-3 worked fine. The problem is that in 1 the description is already a link, so the result ended up being something like '<a href="rev/...">Issue <a href="b.p.o/12345">#12345</a> is now fixed</a>'. With the new solution 1-2 work (the links are added/moved at the end), but it's glitched for case 3. Unless interhg provides a way to limit the replacement only to specific places and/or use different replacements for different places, we will either have to live with these glitches or come up with a proper fix done at the right level. Best Regards, Ezio Melotti
Regards
Antoine.
On Wed, Jan 23, 2013 at 6:16 PM, Ezio Melotti <ezio.melotti@gmail.com> wrote:
How does the above relate to this issue? http://bugs.python.org/issue15919 --Chris
Antoine Pitrou <solipsis@pitrou.net> writes:
I'm not sure if this is related to the upgrade specifically (or just some other hg issue0, but yesterday my Windows 7 buildbot failed an hg pull operation on the 3.x branch and automatically removed it's local repository. Ever since then an attempt to start fresh (with an hg clone) is dying during the clone operation. I'm running the clone operation manually now, and after sitting there for 10 minutes or so at the "adding changesets" stage, it died just like in the buildbot logs ("transaction abort!" followed by "rollback completed"). Nothing was happening on the buildbot during the adding changesets delay so I'm assuming that was all server-side. -- David
Le Wed, 23 Jan 2013 17:37:53 -0500, David Bolen <db3l.net@gmail.com> a écrit :
Given the delay (10 minutes), I'm assuming this has more to do with the connectivity problem Noah is trying to investigate: http://mail.python.org/pipermail/catalog-sig/2013-January/004736.html Regards Antoine.
participants (5)
-
Amaury Forgeot d'Arc -
Antoine Pitrou -
Chris Jerdonek -
David Bolen -
Ezio Melotti