Would somebody using the latest GIT let me know what version numbers are displayed for kate and kwrite (Help/About)?
Thanks.
Darrell
On 06/03/2012 02:57 PM, Darrell Anderson wrote:
Would somebody using the latest GIT let me know what version numbers are displayed for kate and kwrite (Help/About)?
Thanks.
Darrell
Darrell,
Currently rebuilding all on gcc 4.7.0-6 -- I'll have version numbers in an hour or so.
On 06/04/2012 01:20 PM, David C. Rankin wrote:
On 06/03/2012 02:57 PM, Darrell Anderson wrote:
Would somebody using the latest GIT let me know what version numbers are displayed for kate and kwrite (Help/About)?
Thanks.
Darrell
Darrell,
Currently rebuilding all on gcc 4.7.0-6 -- I'll have version numbers in an hour or so.
Darrell,
It's "kwrite 15.0.0" -- what? See:
http://www.3111skyline.com/dl/dt/trinity/ss/kwrite-version.jpg
It's "kwrite 15.0.0" -- what? See:
http://www.3111skyline.com/dl/dt/trinity/ss/kwrite-version.jpg
Okay, thanks. Same here.
I looked at the sources. kate's version number is created by using the system release major number - 1. Or 14 - 1 == 13.
kwrite's version number is created by using the system release major number + 1. Or 14 + 1 == 15.
In the old 3.5.x days the versions were 2.5.x and 4.5.x respectively. Now with us moving to R14, the kate and kwrite version numbers looks way out in left field.
I vote we patch both to eliminate the silly " - 1 " and " + 1 " method. Then both editors will have version numbers the same as the system release. That would look more sensible.
Darrell
On 06/04/2012 11:06 PM, Darrell Anderson wrote:
I vote we patch both to eliminate the silly " - 1 " and " + 1 " method. Then both editors will have version numbers the same as the system release. That would look more sensible.
Darrell
Agreed,
This is a small issue, but I can't see any reason not to get the version numbers the same if we can do it (by hook or crook)
Have you tried a recent gcc 4.7 build on slack to see if there has been any change to the kwrite crash (bug 979) issue. Apparently, there is 1 build on Fedora 17 that works, but all builds on gcc 4.7 on my end have failed with the exact same reproducible bug. Isolating this is way beyond what I can do from a debugging experience level.
Have you tried a recent gcc 4.7 build on slack to see if there has been any change to the kwrite crash (bug 979) issue. Apparently, there is 1 build on Fedora 17 that works, but all builds on gcc 4.7 on my end have failed with the exact same reproducible bug. Isolating this is way beyond what I can do from a debugging experience level.
I haven't tried building with 4.7 because Slackware Current still has 4.7.0. In other words, nothing has changed.
I notice a 4.7.1 release is expected Real Soon Now. I hope Slackware updates respectively. Let me know whether you see any announcements that 4.7.1 is released. I'll then ask the Slackware maintainer to update although he might say no (many other packages have to be rebuilt with a gcc change).
I don't know what magic Francois is using with Fedora 17.
Can you run the TWin4 game, nee KWin4?
Darrell
On 06/05/2012 11:33 AM, Darrell Anderson wrote:
Thanks.
If you can't play please add a confirmation comment in bug report 1014.
Darrell
Will do. I just kicked off the rest of the build, so it will be done by tonight. I'm building everything from tdebase on. So at least I will have a current check on all the build scripts and binaries for testing.
I vote we patch both to eliminate the silly " - 1 " and
" + 1 " method. Then both editors will have version numbers the same as the system release. That would look more sensible.
Agreed,
This is a small issue, but I can't see any reason not to get the version numbers the same if we can do it (by hook or crook)
Here is my proposed patch, which I have tested:
=========================================================== diff -urN tdebase/kate/app/kateapp.cpp tdebase.new/kate/app/kateapp.cpp --- tdebase/kate/app/kateapp.cpp 2012-05-27 18:05:02.000000000 -0500 +++ tdebase.new/kate/app/kateapp.cpp 2012-06-05 11:44:35.000000000 -0500 @@ -131,7 +131,7 @@ */ TQString KateApp::kateVersion (bool fullVersion) { - return fullVersion ? TQString ("%1.%2.%3").arg(KDE::versionMajor() - 1).arg(KDE::versionMinor()).arg(KDE::versionRelease()) + return fullVersion ? TQString ("%1.%2.%3").arg(KDE::versionMajor()).arg(KDE::versionMinor()).arg(KDE::versionRelease()) : TQString ("%1.%2").arg(KDE::versionMajor() - 1).arg(KDE::versionMinor()); }
diff -urN tdebase/kate/app/kwritemain.cpp tdebase.new/kate/app/kwritemain.cpp --- tdebase/kate/app/kwritemain.cpp 2012-05-27 18:05:02.000000000 -0500 +++ tdebase.new/kate/app/kwritemain.cpp 2012-06-05 11:44:03.000000000 -0500 @@ -543,7 +543,7 @@ KLocale::setMainCatalogue("kate"); //lukas: set this to have the kwritepart translated using kate message catalog
// here we go, construct the KWrite version - TQString kWriteVersion = TQString ("%1.%2.%3").arg(KDE::versionMajor() + 1).arg(KDE::versionMinor()).arg(KDE::versionRelease()); + TQString kWriteVersion = TQString ("%1.%2.%3").arg(KDE::versionMajor()).arg(KDE::versionMinor()).arg(KDE::versionRelease());
KAboutData aboutData ( "kwrite", I18N_NOOP("KWrite"), ===========================================================
With the patch both kate and kwrite display a version number of 14.0.0.
Any objections to pushing?
Darrell
On 06/05/2012 12:24 PM, Darrell Anderson wrote:
With the patch both kate and kwrite display a version number of 14.0.0.
Any objections to pushing?
Darrell
Darrell,
With all these version numbers -- Why are the version number not 3.5.14? To me 14.0.0 has no meaning. I know what 3.5.14 is, why not have the full version? Every time I have to dig into the code, I see checks for primary versions of 3 or 4, and then minor numbers of >= 2, etc... No where do I see primary version of 14. I don't know if it makes a difference, but I thought the about dialogs should read the actual version number (eg 3.5.14)
David C. Rankin wrote:
With all these version numbers -- Why are the version number not 3.5.14? To me 14.0.0 has no meaning. I know what 3.5.14 is, why not have the full version? Every time I have to dig into the code, I see checks for primary versions of 3 or 4, and then minor numbers of >= 2, etc... No where do I see primary version of 14. I don't know if it makes a difference, but I thought the about dialogs should read the actual version number (eg 3.5.14)
The version numbers from KDE are based on binary compatibility. This means all 3.5.x versions of KDE are binary compatible and any 3.5.x version can run applications compiled against any 3.4.x (or 3.3.x, 3.2.x, etc.) version of KDE.
However it was decided to drop the binary compatibility requirement in Trinity in favor of ease of maintenance, thus these version numbers are not appropriate anymore.
In the past there were critics written about Trinity, because people were confused about binary compatibility expectations caused by Trinity using confusing version numbers.
This is why a new version numbering system was introduced where the major version number is increased as appropriate, signifying the fact that Trinity does not care about binary compatibility.
In retrospect, all binary incompatible Trinity versions (with KDE 3.5 and 3.x) should have been released with different version numbers.
Even right now, the current version number of the development series is maybe not updated enough, as there are many posts on this mailing list which indicate that even some packagers are not aware when exactly they have to do a full rebuild to avoid binary compatibility issues. I'm not really sure what an appropriate solution for that would be though. (Maybe something like 13.9.x as KDE, as KDE was calling their prerelease 4.0 versions 3.9 as well.)
Best regards, Julius
On 06/07/2012 07:42 AM, Julius Schwartzenberg wrote:
The version numbers from KDE are based on binary compatibility. This means all 3.5.x versions of KDE are binary compatible and any 3.5.x version can run applications compiled against any 3.4.x (or 3.3.x, 3.2.x, etc.) version of KDE.
However it was decided to drop the binary compatibility requirement in Trinity in favor of ease of maintenance, thus these version numbers are not appropriate anymore.
OK,
Thanks Julius. That was the part of the puzzle I was missing. It doesn't matter what the version numbers are as long as we know what they mean. So 14 it is :)
With the patch both kate and kwrite display a version
number of 14.0.0.
Any objections to pushing?
With all these version numbers -- Why are the version number not 3.5.14? To me 14.0.0 has no meaning. I know what 3.5.14 is, why not have the full version? Every time I have to dig into the code, I see checks for primary versions of 3 or 4, and then minor numbers of >= 2, etc... No where do I see primary version of 14. I don't know if it makes a difference, but I thought the about dialogs should read the actual version number (eg 3.5.14)
To me all of the various version numbers are flaky. Change them all to R14.0.0? Then all version numbers become useless because they all match the release number.
Kate/kwrite are the weirdest. Version numbers of 15.0.0 and 13.0.0 respectively look crazy -- not credible. Just like Chrome and Firefox version numbers are not credible.
The calculation method for kate/kwrite worked okay in the previous 3.x.x scheme but blows up with the R14 scheme. My patch retains the calculation method, which then matches the release version. A simple fix. If we want to retain the old version schemes for both apps then we need to reset/hard-code the first and second numbers of the version numbers to 2.5 and 4.5 respectively, and sync the third number with Trinity's R MAJOR number (14). That version scheme makes sense to me too.
I will try to test that idea with a new patch.
Julius explained the new version scheme. Check the mail list archives for further discussion. Many patches in Trinity broke backwards binary compatibility. Further, the new version scheme severs Trinity from KDE 3.5.10 to mold a new identity. Yet like many decisions in life, simple changes suffer from the effects of the "law of unintended consequences." I like the new release version scheme but don't like how that change affects the kate/kwrite versions.
I don't want us wasting time debating version schemes --- we have far more important bugs to resolve if we are ever to release R14. I just want kate and kwrite to have credible version numbers. The slogan "Image is everything" does have merits to a certain degree. :-)
Darrell
To me all of the various version numbers are flaky. Change them all to R14.0.0? Then all version numbers become useless because they all match the release number.
Kate/kwrite are the weirdest. Version numbers of 15.0.0 and 13.0.0 respectively look crazy -- not credible. Just like Chrome and Firefox version numbers are not credible.
The calculation method for kate/kwrite worked okay in the previous 3.x.x scheme but blows up with the R14 scheme. My patch retains the calculation method, which then matches the release version. A simple fix. If we want to retain the old version schemes for both apps then we need to reset/hard-code the first and second numbers of the version numbers to 2.5 and 4.5 respectively, and sync the third number with Trinity's R MAJOR number (14). That version scheme makes sense to me too.
I will try to test that idea with a new patch.
I created and tested a new patch for the kate and kwrite version numbers. I hard-coded the first two numbers and synced the third number to the Trinity release.
With the patch Kate's version number becomes 2.5.14 and kwrite's version number becomes 4.5.14. For those unfamiliar, the respective versions in KDE 3.5.10 were 2.5.10 and 4.5.10. I don't have a 3.5.13 testing environment, but I presume the respective versions there were 2.5.13 and 4.5.13.
The patch retains that consistency and has comments why the new version calculation method is used.
Here is the patch should anybody want to help test or review:
http://humanreadable.nfshost.com/trinity/patches/tdebase-kate-kwrite-version...
Tim, any objections to pushing to GIT?
Darrell