World Wine News

This week, 233 posts consumed 339 K. There were 74 different contributors. 40 (54%) posted more than once. 21 (28%) posted last week too.

The top 5 posters of the week were:

1. 20 posts in 29K by austinenglish at gmail.com (Austin English)
2. 19 posts in 29K by robertshearman at gmail.com (Rob Shearman)
3. 17 posts in 13K by dank at kegel.com (Dan Kegel)
4. 11 posts in 12K by julliard at winehq.org (Alexandre Julliard)
5. 10 posts in 14K by wine at mkarcher.dialup.fu-berlin.de (Michael Karcher)

Welcome to the first World Wine 2009 Newsletter! Okay, I'm sorry, that rhyme just had to be made. 2008 was an exciting year for Wine, from releasing Wine 1.0 to having Photoshop CS2 work nigh flawlessly, (and CS3 installing) to the release of CrossOver Games and CrossOver Chromium. Realistically its likely that 2009 will see Wine 1.2 which will (unrealistically) include flawless DirectX 10 and Wine64 support as well as a fully functioning DIB engine. (Realistically Wine64 may happen. See later in this issue for status of the DIB engine).

What would a new years edition of the WWN be without annual commit statistics? This year provided by Hans Leidekker:

$ for year in {2002..2008}; do \
  count=$( git log | grep ^Date: | grep $year | wc -l ); \
  echo "Number of commits in $year: $count"; \
  done
Number of commits in 2002: 3094
Number of commits in 2003: 3283
Number of commits in 2004: 3851
Number of commits in 2005: 6006
Number of commits in 2006: 8431
Number of commits in 2007: 9532
Number of commits in 2008: 11292

Paul Vriens with a slight modification:

I had a look at these stats as I'm (still) playing around with gitstat
in combination with Wine. The numbers didn't match up so I had a closer
look. 'git log' shows the author date and not the commit date.

So maybe this one is more correct?

$ for year in {2002..2008}; do \
   count=$(git log --pretty='format:%cd' | grep $year | wc -l ); \
   echo "Number of commits in $year: $count"; \
   done
Number of commits in 2002: 3094
Number of commits in 2003: 3283
Number of commits in 2004: 3851
Number of commits in 2005: 6006
Number of commits in 2006: 8404
Number of commits in 2007: 9540
Number of commits in 2008: 11307

Oh and btw, Happy New Year to all !

Eric Pouech brings up a slightly different statistic, unique authors:

digging a bit deeper:
for year in {2002..2008}; do \
count=$( git log | grep -2 -e "^Date.*$year" | grep ^Author: | sed -e
's/<.*\>//' | LC_ALL=C sort -u -f | wc -l ); \
echo "Number of committers $count in year: $year"; \
done
Number of committers 185 in year: 2002
Number of committers 166 in year: 2003
Number of committers 183 in year: 2004
Number of committers 211 in year: 2005
Number of committers 197 in year: 2006
Number of committers 217 in year: 2007
Number of committers 234 in year: 2008

of course, this doesn't take into account the different names used by
the same person (ie there is for example a "frangois gouget" in the
committers' list...)
but this gives a good idea of the variation
A+

Since the early 2000s (long, long ago) the Wine community has maintained a test suite to test the behavior of the Windows APIs. Essentially a developer will learn how a specific function should work, and then write a test to ensure that the function behaves properly in many scenarios. This test can then be applied in both Windows (to confirm that the test does in fact represent how the function behaves in Windows) and in Wine, to make sure that Wine's functions output is the same as Windows'. This form of black box engineering is key to Wine's ability to match the Windows APIs and to prevent regressions.

Fortunately, Wine's test suite has grown very large over the years. Unfortunately many of the tests have been broken on Wine for everybody on the planet except for one man. As it turns out this one man is the only man for whom it counts, Alexandre Julliard, where it has always passed. Everyone else had large number of test failures, which led to regressions in various parts of Wine. In prepration for Wine 1.0, a lot of effort was made to make the tests pass for everyone. These efforts, along with Patchwatcher, have brought the test passing rate up drastically. Now, there are three machines capable of passing the tests! (Alexandre's, Stefan D's, and now Alasdair's!)

The succeeding test run was by Alasdair Sinclair on his Fedora 10 laptop.

It may not be obvious from the linked page what passing means. If you drill down in to the Wine based statistics you'll see that one of the columns has an orange header. In fact this is the only orange header, everything else is red. Orange indicates that there were no failed tests. The header is not yet yellow or green because some tests are skipped.

So there was a lot of press in mid-December about some work Scott Ritchie was doing. Scott Ritchie, an Ubuntu MOTU (Master of the Universe), and avid Wine community member, brought up the idea of integrating Wine into Ubuntu Main.

The initial proposal is a very good read. Scott has very high aspirations for the idea:

Feedback to the idea is mixed. Most Linux/Ubuntu users seem enthusiastic about the idea, however there is some concern from developerss and Ubuntu QA folks about the overall end user experience. Dan Kegel even inspired a similar thread on Wine-Devel. Overall the conversation on Wine-devel was hopeful and included some ideas of how it should be done. One of the more interesting pitches comes from Reece Dunn.

Scott Ritchie has done a lot of work analyzing what is needed to be done to achieve a well integrated Wine-Ubuntu experience. As has been brought up in many conversations of this topic the devil is in the details of the integration. Scott has his own wiki page which links toBetter Wine Integration and Best Wine Integration wiki specifications. There is also a Wine Ubuntu Team that you can join if you are interested in helping out!

For some other takes on these events see ZDNet and Phoronix and PC Mech and Linux Loop .

Now and then the issue of Wine on platforms other than Linux comes up. The WWN has highlighted, and it was highlighted at WineConf that the Wine development community is willing to support other platforms and makes a best effort to do so. The limiting factor is simply developers who use other platforms for regular testing. Below an excerpt from Wine-Devel. David Gerard writes in with a wine-users question:

Austin English, Wine bugs enthusiast, responds:

Kai Blin

Austin has, in a personal interview, confirmed his interest in identifying OpenBSD bugs and recently has submitted a number of patches to fix compilation on non-Linux/x86 machines.

Dan Kegel brought up the idea again that Wine could use a utility similar to DXDiag, and that the implementation of such a utility could make a good student project.

Several developers and Wine D3D experts then gave their opinions. Roderick Colenbrander:

Henri Verbeet:

The next step is to find out specifically which function calls are needed to get this information. Dan Kegel:

Henri again:

This part of the conversation seemed to die off (hopefully a student will pick up on it and run with it). Stefan Dösinger later discussed another way to think of these utilities in Wine.

Massimo Del Fedele has been working hard on a new DIB (Device Independent Bitmap) Engine using bits from Jesse Allen and Huw Davies. Alexandre thus far has been reluctant to use either Jesse or Huw's implementation. It has yet to be seen what Alexandre thinks of this attempt. Massimo has reported some success with real world applications (Something others were having trouble with).

Improvements in Wine's ability to compile for 64-bits and run native Win64 applications have made a number of headlines lately. Many of the major roadblocks have been cleared, and some major benchmarks are starting to be met. Maarten has a number of test applications working and Dan Kegel has a new wiki page for compiling Wine64 yourself. There is also a a 64-bit version of winetest available on the wine test page . Speaking of the test suite, Wine64 does pass a significant number of tests. Dan Kegel:

In the last WWN there was a lot of excitement about Stefan Dösinger's vertex pipeline replacement. However, there is some unfortunate news in this area: Stefan has shelved the new pipeline work. It turns out that the new design did not yield significant performance improvements and in many scenarios actually was slower than the original. Stefan:

*Disclaimer: These lists of changes are automatically generated by information entered into the AppDB. These results are subject to the opinions of the users submitting application reviews. The Wine community does not guarantee that even though an application may be upgraded to 'Gold' or 'Platinum' in this list, that you will have the same experience and would provide a similar rating.