Re: [htdig-dev] boolean terms in phrase - bug?

[Gabriele B. <an...@ti...>]

At 19.34 11/06/2004, Gilles Detillieux wrote:
>OK, I committed Lachlan's patch, which Joe provided in the patch archives,
>so the parser fix is in now.

Thank you Gilles.

>I did discover a little problem, though, while looking through for things
>that may have slipped through the cracks.  The fixes that Lachlan had
>made back in April to get database creation working on Cygwin included
>a couple manual fixes to db/aclocal.m4 and db/configure, to switch in
>the new os_abs.c.win32 file when needed.  Well, those two fixes were
>lost when Gabriele rebuilt these two scripts using autoreconf earlier
>this month.  Now if I understand correctly, the configure script is
>built from configure.in and aclocal.m4, but what builds aclocal.m4 and
>how can we force that to preserve Lachlan's addition?

I am sorry about that. But if we want to easily maintain the configure 
system, we should avoid manually hacking aclocal.m4 files. Usually we 
should modify the acinclude.m4 and regenerate aclocal.m4 through aclocal.

In this case however, I would suggest a different approach, that is to say 
insert a conditional compilation into the os_abs.c file. At the end there 
is the simple patch I'd commit. Let me know what you think and if you have 
the chance to try it, let me know (I don't have any C++ compiler on Windows).

>I suppose we can manually re-add this fix to these two files before
>releasing b6, and deal with the problem properly later, but we really
>need to deal with this.  The problem with all these autotools is that
>it's extremely difficult to keep straight what builds what (and what
>clobbers what!), so how to you make changes permanent?

I would stick with this solution, which I consider more linear (after all I 
think it is a source problem, not an environment detection one).

>Gabriele and Lachlan, I'll throw this in your hands.  You can decide
>between the two of you how you want this fixed for b6 and for 3.2.0 final,
>and how much this should (further) delay the b6 release.

We say: we've done 100, let's do 101 ... I would rather wait until this is 
fixed too and release 3.2.0b6. I think we've done a superb job especially 
in the last week, thanks to the contribution of all of you guys.

I ask you another big favour: if someone of you could please revise the 
Release.html file.

Ciao and thanks,
-Gabriele

--- os_abs.c    2 Feb 2002 18:18:05 -0000       1.2
+++ os_abs.c    12 Jun 2004 07:43:03 -0000
@@ -27,5 +27,9 @@ int
  CDB___os_abspath(path)
         const char *path;
  {
+#if defined (_WIN32) || defined (__MSDOS__) || defined (__DJGPP__) || 
defined (__CYGWIN__)
+       return (path[0] == '/' || path[0] == '\\' || path [1] == ':');
+#else
         return (path[0] == '/');
+#endif
  }