1. 17 Sep, 1998 13 commits
  2. 16 Sep, 1998 6 commits
  3. 15 Sep, 1998 3 commits
  4. 14 Sep, 1998 6 commits
  5. 13 Sep, 1998 1 commit
  6. 12 Sep, 1998 4 commits
  7. 11 Sep, 1998 1 commit
  8. 10 Sep, 1998 6 commits
    • Bill Stoddard's avatar
      These patches have been committed. · 29cdf24b
      Bill Stoddard authored
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82031 13f79535-47bb-0310-9956-ffa450edef68
      29cdf24b
    • Bill Stoddard's avatar
      Remove NT 64 thread limit. · ca8e9b41
      Bill Stoddard authored
      Submitted by:	Ken Parzygnat, Bill Stoddard
      Reviewed by:	Bill Stoddard
      
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82030 13f79535-47bb-0310-9956-ffa450edef68
      ca8e9b41
    • Bill Stoddard's avatar
      Remove NT 64 thread limit. · cc039ef7
      Bill Stoddard authored
      Submitted by: Ken Parzygnat, Bill Stoddard
      Reviewed by: Bill Stoddard
      
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82029 13f79535-47bb-0310-9956-ffa450edef68
      cc039ef7
    • Ralf S. Engelschall's avatar
      Fix possible buffer overflow situation in suexec.c. · a763becc
      Ralf S. Engelschall authored
      PS: The PR#2790 provides a few more fixes for problematic
          things in suexec.c. Because we have to be very carefully here, other
          should review them, too. That's why I commit only some of the fixes from
          this patch.
      
          The other problem we _really_ have to fix is the docroot-check at line
          428. But here the patch from the PR submitter seems to introduce a new
          problem: It accesses cwd[dlen] which can be out of memory bounds. Here
          memory bounds have to checked first.
      
          So, I would appreciate when someothers look at PR#2790, please.  At least
          the docroot-check _has_ to be fixed by us! But correctly, i.e. without
          introducing new problems, of course ;_)
      
      Submitted by: Jeff Stewart <jws@purdue.edu>
      Reviewed by: Ralf S. Engelschall
      PR: 2790
      
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82028 13f79535-47bb-0310-9956-ffa450edef68
      a763becc
    • Ralf S. Engelschall's avatar
      Add some more LIBS for the SCO5 platform which are needed for the already · f766139c
      Ralf S. Engelschall authored
      used -lprot. It's actually a bug in SCO5, of course.
      
      The background is this feedback from R.Record after I replied to him:
      ``Third, i still find the addition of "-lx -ltinfo -lm" to be necessary on
      OpenServer. It's no doubt a bug or at least incorrectly documented, but any
      time -lprot is used you also need -lx and -ltinfo to get nap(), cur_term(),
      setup_term() and others. I've informed our development system guys and this
      may be fixed in a later release but for now these libs are necessary. FYI,
      libx is the SCO Xenix enhancements which have been carried forward since time
      immemorial while libtinfo is the terminfo library. It looks like libprot is
      supposed to setup its own curses stuff but isn't built correctly''
      
      BTW, the following he also mentioned:
      ``First, a kernel bug in OpenServer 5.0.0 and 5.0.2 made it necessary to build
      Apache with USE_FCNTL_SERIALIZED_ACCEPT.  That bug was fixed in 5.0.4 and a
      patch made available for 5.0.0/5.0.2 systems. Further, i have created an
      installation script which detects what release you are installing on and
      whether you need to apply the "Network Maintenance Supplement".  Therefore, i
      can now build Apache without the USE_FCNTL_SERIALIZED_ACCEPT which gets me
      much better performance.  So, as long as you get Apache for OpenServer from
      _me_ (that is, the SCO Skunkware CD-ROM or web site), you're ok. If you build
      it yourself, you'd better know whether you need to apply the appropriate
      support supplement.''
      
      But I still think it's not really good to remove the
      USE_FCNTL_SERIALIZED_ACCEPT. The only correct solution would be to dispatch
      for 5.0.0 and 5.0.2 and avoid it only for >= 5.0.4.
      
      Submitted by: Ronald Record <rr@sco.com>
      PR: 2533
      
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82027 13f79535-47bb-0310-9956-ffa450edef68
      f766139c
    • Ralf S. Engelschall's avatar
      Rename some references in the docs from the old "OSF1" term to the official · c235c045
      Ralf S. Engelschall authored
      term "Digital UNIX".
      
      Submitted by: Robert Urban <urban@rto.dec.com>
      
      
      git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@82026 13f79535-47bb-0310-9956-ffa450edef68
      c235c045