Started 1 yr 10 mo ago

Aborted Build #10 (Jul 8, 2010 1:16:24 PM)

Changes
  1. bring over likely libz fix from head (detail)
  2. copy from head to address problems Olaf reported to me
    This does reduce test coverage in release, in that head
    doesn't test stuff that got deleted in head. Should be ok.
    (stuff like making sure the cloned headers line up) (detail)
  3. make it build on FreeBSD 8.0 for AMD64, too (detail)
  4. update generated HTML doc to RC5 (detail)
  5. update generated HTML doc to RC5 (detail)
  6. acceptable hack for release branch, don't put in head (detail)
  7. version string after RC5 (detail)
  8. update generated HTML doc to RC5 (detail)
  9. whitespace only (detail)
  10. port from head: don't make PPC_LINUX executables position independent, as it breaks debugging (we do use this on Solaris actually, but the odds of such a bug being portable aren't high) (detail)
  11. Port unfortunate fix from head, unless/until a better fix is found.
    The program:

    MODULE Main;

    PROCEDURE F1() =

        PROCEDURE NestedUnused1() =
        BEGIN
        END NestedUnused1;

    BEGIN
      IF FALSE THEN
        NestedUnused1();
      END;
    END F1;

    BEGIN
      F1();
    END Main.


    fails to link with:
    Undefined symbols:
      "_Main__F1__NestedUnused1.496", referenced from:
          _L_1 in Main.mo


    unless we do:
      flag_unit_at_a_time = 0; in parse.c (detail)
  12. manage version like in head, should fix the Solaris problem (detail)
  13. try again (detail)
  14. adjust comment: spell out 'pre' means (detail)
  15. remove whitespace from ends of lines (detail)
  16. some means to generate HTML, PS and PDF documents from m3-defn-complete.tex
    along with the generated documents for WWW access
    old versions (two directories above) not removed yet (detail)
  17. update generated HTML doc to RC5 (detail)
  18. add test case to release branch -- looks broken on Solaris, HP-UX, Interix, Cygwin -- timezone parameter isn't handled correctly in Date.FromTime (detail)
  19. forgot to add this: copied from head; the point is to cut the dependency on m3core/Utime, so that release can be 'upgraded' from head (detail)
  20. copy from head: fixes for noresolve, abandon Target.EOL (just use Wr.EOL),
      mark config files as obsolete (detail)
  21. prepare for final release build (detail)
  22. copy from head so it compiles (detail)
  23. update line information in error message (detail)
  24. ignore 'Illegal instruction', too (detail)
  25. update some release information (detail)
  26. merges corrections from trunk (detail)
  27. copy from head: fixes for noresolve, abandon Target.EOL (just use Wr.EOL),
      mark config files as obsolete (detail)
  28. move _DARWIN_FEATURE_64_ONLY_BIT_INODE earlier, so that it will work
    (This only affects ARM_DARWIN I believe, and maybe not even it.) (detail)
  29. some minor stuff from head (detail)
  30. correct and augment mentions of 64bit LONGINT on NT386 -- it is not in this release, but it is already implemented for the next; add two other AMD64 targets explicitly -- Darwin and FreeBSD (detail)
  31. file m3core.h was added on branch release_branch_cm3_5_8 on 2010-05-09 07:33:23 +0000 (detail)
  32. if SYSTEM_LIBS contains POSTGRES95 like m3-db/postgres95/src/m3makefile (detail)
  33. file version.quake was added on branch release_branch_cm3_5_8 on 2010-04-13 11:31:10 +0000 (detail)
  34. fix newlines, Sun cc complains (detail)
  35. try again from the start (detail)
  36. one more RC5 change (detail)
  37. add post-RC5 link (detail)
  38. merge patch from head (detail)
  39. remove CR at EOL (detail)
  40. Fix from Mika Nystrom for RECORD fields (detail)
  41. conditional ignore ssh unavailability (detail)
  42. fix I386_DARWIN (detail)
  43. bring over the rest of head (stamping version=branch) (detail)
  44. go back a version, the problem is bc (detail)
  45. try to make it work, if I fail, ok (detail)
  46. turn off "pre" optimization, which crashes compiling
      m3-tools/cvsup/server/FSServer.m3 (detail)
  47. fix syntax error
    reduce duplication
      no need to have foostderr and foostdout that are identical, just have foo
    add spaces because on my Mac "[]" looks like a box while
    "[ ]" looks clearly like two characters
    remove one tab; tabs are bad because they do not render consistently,
      sometimes they are 8 spaces, sometimes 4, could be anything really (detail)
  48. fix version for RC5 again (detail)
  49. regenerate PS and PDF documents (detail)
  50. remove whitespace from end of line (detail)
  51. copy from head: fixes for noresolve, abandon Target.EOL (just use Wr.EOL),
      mark config files as obsolete (detail)
  52. file CoffTime.c was added on branch release_branch_cm3_5_8 on 2010-05-10 08:42:09 +0000 (detail)
  53. hidden files are generally not worth using, use foo.temp instead of .foo (detail)
  54. file Main.m3 was added on branch release_branch_cm3_5_8 on 2010-04-15 11:08:09 +0000 (detail)
  55. update link to language definition (detail)
  56. don't run bc on Solaris (based on name=SunOS) (detail)
  57. add checksums display (detail)
  58. copy from head: fixes for noresolve, abandon Target.EOL (just use Wr.EOL),
      mark config files as obsolete (detail)
  59. small partial merge from head to fix problem I hit building SOLsun
    no need to move tag, can be in near future release if there is one

    if host != target, we try to run the host PklFonts (which is bogus actually?
       I put this in in the first place, without it, cross builds try
       to run something that won't run)

    But if host or target are both solsun or solgnu, then go ahead
       and run what we just built, it isn't really a cross build.

       There are other combinations like this and we should come
        up with some abstraction:
          CanHostRunTarget(target)

      (e.g. I386_DARWIN and AMD64_DARWIN can run PPC_DARWIN
            PPC_DARWIN can sometimes run I386_DARWIN and AMD64_DARWIN (if OS is x86)
            CYGWIN can run NT and MINGW and MSDOS
            NT and MINGW can run each other
            NT can sometimes run CYGWIN
            MSDOS can sometimes run NT/MINGW/CYGWIN
            FOO64_FOO can often run FOO32_FOO (ditto AMD64/I386)
            FOO32 can sometimes run FOO64 (if FOO32 cm3 running on FOO64 OS) (detail)
  60. adaptions for opencsw environment from head (detail)
  61. more ignore texts for Linux (detail)
  62. fix for 64bit Darwin (AMD64_DARWIN, PPC64_DARWIN, though
      PPC64_DARWIN has other problems in the release branch)

    Note that head went through a period with safer copying wrappers
    and now pushes C up a level to end up generally removing code,
    and all the Usysdep.i3 files are gone now, in head.

    This form of the fix was chosen for its focus on specific broken targets.
    It is in a Darwin specific file and changing INTEGER to int doesn't
      affect 32bit targets. (detail)
  63. prepare for final release build (detail)
  64. fix: list .deb and .msi packages with the target-specific packages (detail)
  65. add more SPARC64 targets (detail)
  66. copy from head, so that release can be built by "upgrading" from head
    (head m3core has cut down/removed Utime for portability, and it
    was used here)) (detail)
  67. support SHIPRC=1 besides 'y' and 'yes' (detail)
  68. from head: use correct variable name (basic problem of no good Unix editor so I generally edit on Windows and copy back/forth to test..tedious.. (detail)
  69. simplify like in head, so fix Solaris (detail)
  70. split up tar and gzip instead of using GNU tar's -z flag (for building .deb files for Solaris, wierd, I know) (detail)
  71. add file to support what cvsup.quake is trying to do, mark itself with a branch, but that I broke, but that I'm scared to fix in a way that resembles 'putting it back' because running sh/awk is just too fragile we find repeatedly (detail)
  72. file m3makefile was added on branch release_branch_cm3_5_8 on 2010-04-15 11:08:09 +0000 (detail)
  73. fix simple copy/paste bugs (detail)
  74. change C++ slashslash comment to C slashstar comment (detail)
  75. TARGET must be back-substituted later, after INSTALL_ROOT, Olaf
    correctly points out. (detail)
  76. if SYSTEM_LIBS contains ODBC like m3-db/odbc/src/m3makefile (detail)
  77. add a simple workaround to ignore some texts and lines for the r-tests (detail)
  78. allow CM3CG to be preset (detail)

Started by user wagner