| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
|
|
|
|
| |
This makes gdl version status to be included in the report, instead of ignored.
|
|
|
|
| |
String exceptions have been deprecated a long time ago.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
version for a pkg.
In 2.27.91 the craze of using nonsensical version strings continues (before we had
a gucharmap-2.27.0-git) in the form of a gnome-media-2.27.90.fix release, which also
happened to end up in releng release versions list for some reason.
If a package parsing fails, don't append it to the list.
|
|
|
|
|
|
|
|
|
|
| |
packages in release_packages
Fix crash when vuntz latest packages walking script doesn't have a version for some package for some reason,
while it does exist in the GNOME releng official release versions listing.
Currently versions-2.28 file doesn't have the latest gucharmap release version for some reason and is
completely omitted, causing the problem in this script too.
|
|
|
|
|
| |
Now one can for example simply do print(latest_packages) to print out the whole
array for debugging.
|
|
|
|
|
|
|
|
| |
Perl bindings don't have a 'latest' version field because they are
hosted on different infrastructure from the rest of GNOME. This commit
cleans up the way these are handled (checks the language of the binding
rather than depending on the package name, which broke with the
introduction of the Pango module).
|
|
|
|
|
| |
Reworded a bit so I can understand what it says myself and added a missing
linebreak, so the descriptions aren't bunched together.
|
|
|
|
|
|
|
|
|
|
|
| |
Traceback (most recent call last):
File "./gentoo-bumpchecker.py", line 49, in <module>
gnome_output.Output(comparison_result_packages, True).generate_html()
File "modules/gnome_output.py", line 63, in generate_html
lines.append('<title>Gnome ' + clioptions_module.Options().get_arguments().release_numer + ' Progress</title>')
AttributeError: Values instance has no attribute 'release_numer'
Fixed typo to release_number
|
|
|
|
| |
Signed-off-by: Gilles Dartiguelongue <eva@gentoo.org>
|
| |
|
|
|
|
|
| |
Use pastel colors, pretty borders, and smaller font sizes ;p
Also, reduce column width
|
|
|
|
|
|
|
| |
Shortens it from 17 lines to 1 line by using str.split and str.join instead
of the unwieldy hand-rolled dot finding string walking method.
There is no worry for performance here as version numbers won't be long
for string walking till the end being of concern.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Packages such as gconf, glade and libxmlpp almost always reported compliance
with dark green color, even though latest available and/or official release
version were newer than portage version.
The version comparison is done via the portage module best() function, which
takes a list of package strings (package name including version number) and
returns the newest of it.
best_version_test() takes Package.name_plus_version strings, but those weren't
fixed up according to the special_cases name mapping in
Package.handle_special_cases(), and portage gets package strings such as
"gconf-2.22.0" and "GConf-2.23.2" for picking the best version of it, and it
considers lower case gconf-2.22.0 newer than GConf-2.23.2, probably due to some
specific logic or non-handling of different string package names.
To fix it, instead of uglifying handle_special_cases() with setting of
name_plus_version and name_plus_version_plus_revision, convert them into
read-only attributes that get constructed on the fly -- properties with only a
getter.
For readability decorators are used, making >=python-2.4 a requirement, with
which I see no problem in a developer tool.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
* By default, no overlay is scanned
* --all-overlays can be used to scan all overlays in addition to PORTDIR
(current behaviour)
* -a/--overlay can be used to specify one overlay (or more than one if
specified multiple times)
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Here's a first stab at a patch that replaces depchecker's network bound
slow GNOME FTP scraping with parsing of "versions" formatted files.
Earlier we could have gotten the release versions quickly this way as
well, cutting the time spent on scraping to half, but now vuntz made
server-size hourly versions file of latest versions for each gnome
version and we can parse that.
The URL is temporary and will change later.
This patch is for feedback purposes. To my dirty code and especially for
feedback to pass on to vuntz about the files that are in that /tmp
directory of his currently.
I've linked up my runs at http://dev.gentoo.org/~leio/gnome/ for
comparison of the FTP scraping and versions file results. the sorting
seems to be different a bit, but I don't have time to get that to match
right now. Heavy improvements possible, etc. I just did a first working
versions to see if the upstream file works out and will go back to GNOME
2.22 stuff. So, anyone else can improve on top of that - I won't revisit
it for some days.
For example there could be --stable and --unstable switches (latest
upstream stable release cycle, latest upstream development release
cycle). See
http://www.gnome.org/~vuntz/tmp/versions/
I don't think this should be pushed before the upstream URL settles and
we compare if the results are not worse, etc.
I think I also have no proper --nextrev handling in regards to the file
we need to use then.
commit ad04a2ee64ffbd8237c2dbd6847751f0a0caa482
Author: Mart Raudsepp <leio@gentoo.org>
Date: Wed Mar 19 00:41:46 2008 +0200
Use versions files generated in upstream server in an hourly basis instead of slow FTP scraping
Signed-off-by: Daniel Gryniewicz <dang@gentoo.org>
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|