Den mån 18 maj 2026 kl 20:21 skrev Branko Čibej <[email protected]>: > > On 18. 5. 26 20:16, Branko Čibej wrote: > > On 18. 5. 26 20:06, Daniel Sahlberg wrote: > > Den mån 18 maj 2026 kl 19:34 skrev Evgeny Kotkov via dev > <[email protected]>: > > Hi all, > > We have a number of JavaHL-related nominations in 1.15.x/STATUS: > > ``` > * r1930973 > Fix test failures of JavaHL with Java 25 on Windows due to that deleting a > file with readonly flag on Windows fails since Java 25. > Justification: > Allow JavaHL test to succeed on Windows since Java 25 > Votes: > +0: hartmannathan (cannot test; Windows testers needed please) > +1: jun66j5 > > * r1933796 > Silence JavaHL compiler warnings about missing bootstrap classpath. > Justification: > We should use the cross-compilation option supported by every JDK since > version 9, unless we're actually building with JDK 1.8. > Votes: > +1: brane > +0(concept): dsahlberg, LGTM have not tested under JDK 1.8 > > * r1933874 > Silence the dangling-doc-comment warnings in JavaHL. > Justification: > Unnecessary warnings hide actual problems. > Votes: > +1: brane, dsahlberg > > * r1933877 > Silence all deprecation warnings from the org.tigris JavaHL package. > Justification: > Unnecessary warnings hide actual problems. > Depends: > Votes: > +1: brane, dsahlberg > > * r1934313 > Include 1.15-specific APIs in JavaHL. > Justification: > We should try to keep our bindings up to date. > Votes: > +1: brane, dsahlberg > ``` > > Our rule for bindings-related nominations is that they can be backported with > one +1 and at least one +0, provided that they don't affect the build system. > > While some of these commits touch `build/ac-macros/java.m4`, the changes are > isolated to a specific branch of the `if test -z "$JAVAC_FLAGS"` condition. > Since this is an isolated and a bindings-related part of the code, I would > say the relaxed voting rules also apply here. > > With all that in mind, I plan to mark the above nominations as approved. > Please let me know if there are any objections. > > I also thought about this but I forgot to send it to the list. +1 from me. > > > > It would seem that the backport script stripped away the Depends: versions. > > > Yup. r1934330. That's not good.
Looks like the script[1] expects: [[[ Depends: rNNNN ]]] (ie. in a single line). There is already code to manage both [[[ Branch: [url] ]]] and [[[ Branch: [url] ]]] so it should be fairly simple to extend the same concept to Depends. Unfortunately I'm out of time for today so it have to be another day. Cheers, /Daniel [1] https://svn.apache.org/viewvc/subversion/trunk/tools/dist/backport/status.py?revision=1930598&view=markup#l450
