PING: [PATCH FOR REVIEW] System Zlib Support (original) (raw)

Andrew Hughes ahughes at redhat.com
Wed Aug 15 14:20:21 UTC 2012

----- Original Message -----

Yes, you are good to go.

Thanks! Pushed: http://hg.openjdk.java.net/jdk8/tl/jdk/rev/35e024c6a62c

Thanks! -Sherman >> On 8/14/2012 6:13 AM, Andrew Hughes wrote: > ----- Original Message ----- >> On 14/08/2012 13:36, Andrew Hughes wrote: >>> : >>> >>> Do you have a static libz.a installed (from zlib1g-dev) and all >>> dependent static libraries? >> I think you're right about the static linking but this was just a >> quick >> test to see if SYSTEMZLIB=true worked with everything else as >> default. >> > Ah right. > >> pack200 has C++ so that explains why we see it there and not when >> building libzip. >> > Right. BTW, it appears gcc 4.5 and later have a new option > -static-libstdc++. > I don't know if that would work better than the options currently > used. > > http://gcc.gnu.org/gcc-4.5/changes.html > >> $ dpkg -s zlib1g-dev >> Package: zlib1g-dev >> Status: install ok installed >> Multi-Arch: same >> Priority: optional >> Section: libdevel >> Installed-Size: 366 >> Maintainer: Ubuntu Developers >> <ubuntu-devel-discuss at lists.ubuntu.com> >> Architecture: i386 >> Source: zlib >> Version: 1:1.2.3.4.dfsg-3ubuntu4 >> Provides: libz-dev >> Depends: zlib1g (= 1:1.2.3.4.dfsg-3ubuntu4), libc6-dev | libc-dev >> Conflicts: zlib1-dev >> Description: compression library - development >> zlib is a library implementing the deflate compression method >> found >> in gzip and PKZIP. This package includes the development >> support >> files. >> Homepage: http://zlib.net/ >> Original-Maintainer: Mark Brown <broonie at debian.org> >> >> $ apt-file search libz.a >> lib64z1-dev: /usr/lib64/libz.a >> zlib1g-dev: /usr/lib/i386-linux-gnu/libz.a >> >> $ ls -l /usr/lib/i386-linux-gnu/libz.a >> -rw-r--r-- 1 root root 98772 Nov 10 2011 >> /usr/lib/i386-linux-gnu/libz.a >> >>> I suspect you'll get the same failure with or without my patch >>> (though you'll probably need the change >>> to defines.h to get this far...) >> SYSTEMZLIB is currently Mac only so I wouldn't expect to have got >> very >> far without your changes. > Yes, the defines.h change I refer to is the one that removes the > Mac > clause. It fails earlier without that change (I've been hitting it > on most builds now I have SYSTEMZLIB set), so the patch is an > improvement, even if it doesn't work in all cases. > >>> I'd guess you have STATICCXX set to true (the default >>> apparently): >> I didn't specify any other build options so it's using the >> default. >> > An odd default, though I can understand why, given OpenJDK's > history. > >> BTW: I should mention that I don't have any issues with the patch >> proposed as it works as it does now because SYSTEMZLIB is false. >> It >> may >> be that there is follow-up to allow SYSTEMZLIB=true and static >> linking >> to work together. >> > Can I take that as it's good to push? :-D > >> -Alan. >> > Thanks, >

Andrew :)

Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com)

PGP Key: 248BDC07 (https://keys.indymedia.org/) Fingerprint = EC5A 1F5E C0AD 1D15 8F1F 8F91 3B96 A578 248B DC07

More information about the core-libs-dev mailing list