tags 426715 + wontfix close 426715 thanks Mika S Marjamäki wrote: > In libxcb 1.0-1.2 there was a patch which allowed many broken > applications (e.g. java) to run with newer libxcb: > > - - - > libxcb (1.0-1.2) experimental; urgency=low > > * Add 100_sloppy_lock.diff which allows a workaround for programs > that throw > lock errors. You can set the environment variable > "LIBXCB_SLOPPY_LOCK" > when you run that program and xcb won't exit when it generates a > locking > error. This will allow java to run. > > -- David Nusinow <[EMAIL PROTECTED]> Sun, 08 Apr 2007 11:50:13 -0400 > - - - > > For some reason this rather useful patch is missing from libxcb 1.0.3, > which is currently in unstable.
Quoting a previous message I sent to someone on this topic: On Wed, May 30, 2007 at 02:33:45PM -0700, Josh Triplett wrote: > We do not plan to upload libX11 with Xlib/XCB to unstable until we have some > sort of satisfactory solution for some of the widely-used programs with broken > locking, including Java. (Emphasis on "some"; many of the problems will still > get sorted out in unstable.) If you need to run programs with broken locking > before Sun addresses that problem or we find some other solution, feel free to > run libX11 from unstable and you will not encounter those problems. > > We do not plan to include the patch from that NMU (which occurred without our > knowledge), and the upload to unstable intentionally omitted this patch. This > patch will not go upstream into XCB. No amount of telling people "temporary > workaround" will prevent this from becoming a semi-permanent solution, so we > want to reject it from the start. Almost any use of this workaround will lead > to undefined and likely buggy behavior in XCB or Xlib/XCB. We need people to > pressure the authors of broken software to fix it, and we can't do that when > the easier solution just requires exporting an environment variable. Quoting David Nusinow: > I just want to chime in that even though I did the NMU and included the > patch, I back the XCB maintainers and their plan. I wanted to provide an > escape hatch for people using things in experimental who needed to run java > apps (a group that includes myself) but also do so in a way that allowed us > to catch the application bugs. The patch is a stopgap solution and we need > a proper one before building libx11 against Xlib/XCB and shipping it > generally. jcristau, christoph4 and I managed to track down the problem with Sun Java 1.5 and 1.6, and we have a workaround now. Expect a mail about this this shortly, and a fix in unstable soon. - Josh Triplett
signature.asc
Description: OpenPGP digital signature
