| tdaitx | doko: took me a while to get my qemu updated, test the whole stuff, retest with patches applied, various LaF/swing/gtk settings, screenshots, and bug reports, but I believe that while upstream has fixed some issues (upcoming openjdk 11.0.4) the GTK2 experience is way better and seems to be working fine on bionic | 03:32 |
|---|---|---|
| tdaitx | should we maybe revert openjdk-11 in bionic to rely on gtk2 by default? gtk+2.0 has about 38 reverse (b)deps in Bionic (and 37 in Disco) | 03:32 |
| doko | tdaitx: sure, we can. not sure about disco yet, because we will get these updates for disco. should be optional | 06:57 |
| doko | tdaitx: what exactly is this android build tools problem in disco? | 16:02 |
| tdaitx | doko: jh_build is no longer respecting the --release flag for some reason, I believe it is related to the changes in javatools 0.72.3 | 16:09 |
| doko | ahh, the perl conversion | 16:09 |
| tdaitx | yeah, when they moved from bash to perl | 16:10 |
| tdaitx | it is setting -source 7 since then | 16:10 |
| tdaitx | on bionic: | 16:11 |
| tdaitx | jh_build --javacopts="--release 8 -sourcepath src/main/java" --no-javadoc \ | 16:11 |
| tdaitx | --main=com.android.apksigner.ApkSignerTool \ | 16:11 |
| tdaitx | apksigner.jar src/apksigner/java | 16:11 |
| tdaitx | find src/apksigner/java -name *.java -and -type f -print0 | xargs -s 512000 -0 /usr/lib/jvm/default-java/bin/javac -g -cp /usr/share/java/apksig.jar:debian/_jh_build.apksigner -d debian/_jh_build.apksigner -encoding ISO8859-1 --release 8 -sourcepath src/main/java | 16:11 |
| tdaitx | on disco: | 16:11 |
| tdaitx | jh_build --javacopts="--release 8 -sourcepath src/main/java" --no-javadoc \ | 16:11 |
| tdaitx | --main=com.android.apksigner.ApkSignerTool \ | 16:11 |
| tdaitx | apksigner.jar src/apksigner/java | 16:11 |
| tdaitx | jh_build: find src/apksigner/java -name '*.java' -and -type f -print0 | xargs -s 512000 -0 /usr/lib/jvm/default-java/bin/javac -g -cp /usr/share/java/apksig.jar:debian/_jh_build.apksigner -d debian/_jh_build.apksigner -encoding ISO8859-1 -source 1.7 -target 1.7 | 16:11 |
| doko | tdaitx: but it works in unstable | 16:13 |
| tdaitx | I got an email from Bart Martens that my proposed update in debian failed to build | 16:14 |
| tdaitx | with basically the same error as disco | 16:14 |
| doko | and setting source and target instead of release? | 16:15 |
| tdaitx | back then I only build it locally on bionic, it worked so I uploaded to bionic, cosmic, disco, and debian (mentors)... I was not expecting the problem in disco | 16:16 |
| tdaitx | we shouldn't really be using -source -target with openjdk-11 | 16:16 |
| tdaitx | if any package fails to build with that, it is because it is probably a package that is made to work with (at most) openjdk-8 | 16:16 |
| tdaitx | if any package fails to build with --release 8, it is because it is probably a package that is made to work with (at most) openjdk-8 | 16:17 |
| doko | the logic is wrong | 16:19 |
| doko | if (not _has_java_option(\@javac_opts, '--release') and not _has_java_option(\@javac_opts, '-source')) { | 16:19 |
| doko | # If neither --release nor -source is set, then set -source (and -target if also absent) | 16:19 |
| doko | if (not _has_java_option(\@javac_opts, '-target')) { | 16:19 |
| doko | push(@javac_opts, '-source', DEFAULT_JAVA_RELEASE, '-target', DEFAULT_JAVA_RELEASE); | 16:19 |
| doko | it defaults to DEFAULT_JAVA_RELEASE, not the --release value | 16:19 |
| tdaitx | yeah, but I did set the --release, it should not be using -source -target there | 16:21 |
| tdaitx | I didn't look into that, just saw that this javatools change is what probably broke the build with --release | 16:22 |
| doko | tdaitx: what should it do? just pass --release 8, and not adding -source 1.7 -target 1.7? | 16:24 |
| tdaitx | yeah | 16:24 |
| tdaitx | from a quick look over that patch I have to say that I don't understand why it is even going into the if clause to set default | 16:26 |
| tdaitx | it should only go into that if neither --release and -source are set, but in the FTBFS --release is actually set | 16:26 |
| tdaitx | s/and/nor/ | 16:27 |
| tdaitx | doko: look at this: | 16:29 |
| tdaitx | +# Space-separated list of options | 16:29 |
| tdaitx | +'javacopts|o=s' => sub { @javadoc_opts = split(' ', $_[1])}, | 16:29 |
| tdaitx | +'javadoc-opts=s' => sub { @javadoc_opts = split(' ', $_[1])}, | 16:29 |
| tdaitx | it is not parsing the options correctly, there is a typo there | 16:29 |
| tdaitx | the sub call should be setting the javac_opts array: | 16:31 |
| tdaitx | - 'javacopts|o=s' => sub { @javadoc_opts = split(' ', $_[1])}, | 16:31 |
| tdaitx | + 'javacopts|o=s' => sub { @javac_opts = split(' ', $_[1])}, | 16:31 |
| doko | found it | 16:33 |
| doko | yep | 16:33 |
| doko | tdaitx: gtk2 for disco seems to be fine as well, no plan yet for demotion | 17:22 |
| tdaitx | doko: ok, I was going to test a build that depended on gtk2 for bionic and gtk3 for disco+ (similar to openjdk-8 xenial/bionic), should I default to gtk2 as a dependency then? do we want to have it as an alternate dependency default to gtk2? | 17:24 |
| doko | yes, the latter please | 17:26 |
| doko | I just confirmed that gtk3 support will still take some time | 17:27 |
| tdaitx | doko: ok, so depends "libgtk2.0-0 | libgtk-3-0", but what about b-dep? | 17:28 |
| doko | tdaitx: what did you remove? ;p | 17:29 |
| tdaitx | well building with libgtk-3-dev allows us to run with gtk2 (since it dinamically loaded) | 17:29 |
| doko | b-d should be the default as well | 17:29 |
| doko | I didn't look at the build system. if it has consequences | 17:30 |
| tdaitx | I never tested if b-deps on libgtk2.0-dev works the same way (it should, but) | 17:30 |
| tdaitx | ok, I had some hope you had looked at this back on the openjdk-8 change =) | 17:31 |
| tdaitx | in that case I will check how they use the headers | 17:31 |
| tdaitx | doko: I copied android-platform-tools-apksig from our ppa (bionic) into cosmic-proposed, it needs approval... it got rejected earlier this week, I assume because disco failed(?) but the bionic copy was approved later o_O | 17:34 |
| doko | tdaitx: accepted | 17:35 |
| tdaitx | btw, android-platform-tools-apksig rebuild ok on disco after the javatools fix, tks doko! | 17:36 |
| tdaitx | doko: the upstream bug report for "defaulting to GTK3 instead of GTK2" mentions that the change should be coordinated with javafx | 17:48 |
| tdaitx | https://bugs.openjdk.java.net/browse/JDK-8198649 | 17:49 |
| tdaitx | javafx was tracked in https://bugs.openjdk.java.net/browse/JDK-8198654 and commited to http://hg.openjdk.java.net/openjfx/jfx-dev/rt/rev/3e6c9b6db694 | 17:50 |
| tdaitx | so we might want to update javafx as well and revert/update this: http://hg.openjdk.java.net/openjfx/jfx-dev/rt/rev/3e6c9b6db694 | 17:50 |
| tdaitx | interesting, gtk dev packages are not actually required for building openjdk-11, I removed that build dependency and set the runtime alternatives for jre... uploading for a test build in a ppa now | 19:26 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!