On Wed, Jun 29, 2016 at 12:40 PM, Mario Torre <neug...@redhat.com> wrote: > Ping?
Ping Ping? > Cheers, > Mario > > On Thu, Jun 16, 2016 at 6:47 PM, Mario Torre <neug...@redhat.com> wrote: >> On Wed, Jun 1, 2016 at 1:52 PM, Semyon Sadetsky >> <semyon.sadet...@oracle.com> wrote: >>> >>> >>> On 6/1/2016 2:39 PM, Mario Torre wrote: >>>> >>>> On Wed, Jun 1, 2016 at 1:23 PM, Semyon Sadetsky >>>> <semyon.sadet...@oracle.com> wrote: >>>>> >>>>> I ran JPRT build. It seems that the build server does not have the >>>>> requested >>>>> header: >>>>> >>>>> >>>>> /opt/jprt/T/P1/102200.ssadetsk/s/jdk/src/java.desktop/unix/native/libawt_xawt/awt/awt_Robot.c:40:39: >>>>> fatal error: X11/extensions/Xcomposite.h: No such file or directory >>>>> #include <X11/extensions/Xcomposite.h> >>>>> >>>>> someone need to install Xcomposite library there >>>>> >>>> Should this be made optional? >>>> >>>> I can do some dlsym hack if necessary (I would prefer to avoid that >>>> though). >>> >>> Apparently dlsym is the shortest way to fix that issue and you don't need to >>> modify the makefile in that case. >> >> Well, to be honest, I think Xcomposite should just be a requirement, I >> believe all the systems that OpenJDK 9 targets as "supported" have >> that library, it just needs to be installed. >> >> Anyway, this is the patch with the dl-stuff hacks: >> >> http://cr.openjdk.java.net/~neugens/8150954/webrev.06/ >> >> The only question I have is where to unload the library, I gave a look >> at what the XToolkit code does with GTK, and it seems that it's just >> never unloaded (there's a method to unload the library, but looks like >> it's never used anywhere, except, of course, if some functions fail to >> load in the first place, which is what I also do here). I think this >> is not a real problem though, so rather than making the whole code a >> lot more complex I would keep it that way, YMMV. >> >> I believe I need again two approvals now. >> >> Cheers, >> Mario
