Re: Compiling gawk on Tandem NSK/OSS

[Aharon Robbins]

Thanks for this note.  As to snprintf, there's a replacement version
in missing_d.  Can you tell why configure isn't finding it?

Thanks for the other patches, I'll work on getting them into CVS sometime
this week.

Thanks!

Arnold

In article <address@hidden> you write:
>-=-=-=-=-=-
>
>(Sorry for the delay, I finally got around to trying this...)
>
>Aharon Robbins wrote:
>> Greetings. Now that autoconf 2.61 will define _TANDEM_SOURCE for us,
>> I have incorporated a modified version of this patch that should let
>> gawk build out of the box on a Tandem system. The patches will show
>> up in the gawk CVS archive shortly.
>
>Well. That was fun. CVS is not buildable. NSK (at least my revision) 
>does not have snprintf, and CVS builtin.c blithely assumes it. I *think* 
>I would get a successful link if that was fixed, although I also had to 
>apply the attached patch.
>
>> I didn't do anything about the library; if you can add a shell snippet
>> to configure.ac to do the trick and send me a patch relative to the current
>> CVS, that will help.
>
>I didn't check this part with CVS strictly OOTB, this trivial little 
>change checks libzrldsrl for dlopen if it isn't in libdl. With this, 
>configure sets LIBS correctly. A cursory check (try the link by hand - 
>which fails, see above - without -lzrldsrl) complains about dlopen, 
>dlerr, and dlsym as well as snprintf, so it seems it is still needed.
>
>==== BEGIN patch
>Index: configure.ac
>===================================================================
>RCS file: /sources/gawk/gawk-stable/configure.ac,v
>retrieving revision 1.6
>diff -u -r1.6 configure.ac
>--- configure.ac        2 Feb 2007 08:47:18 -0000       1.6
>+++ configure.ac        8 Feb 2007 20:35:41 -0000
>@@ -282,7 +282,8 @@
>
>         # Check this separately. Some systems have dlopen
>         # in libc. Notably freebsd and cygwin.
>-       AC_CHECK_LIB(dl, dlopen)
>+       # HP-NSK has it in zrldsrl
>+       AC_SEARCH_LIBS(dlopen, dl zrldsrl)
>  ])
>
>  dnl check for how to use getpgrp
>==== END patch
>
>-- 
>Matthew
>"Two IIRC's must make a right" -- Larry Hall (paraphrased)
>
>-=-=-=-=-=-
>
>Index: regex_internal.h
>===================================================================
>RCS file: /sources/gawk/gawk-stable/regex_internal.h,v
>retrieving revision 1.3
>diff -u -r1.3 regex_internal.h
>--- regex_internal.h    14 Jan 2007 20:41:02 -0000      1.3
>+++ regex_internal.h    8 Feb 2007 21:53:00 -0000
>@@ -437,10 +437,9 @@
> #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
> #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
>
>-#include <alloca.h>
>-
> #ifndef _LIBC
> # if HAVE_ALLOCA
>+#  include <alloca.h>
> /* The OS usually guarantees only one guard page at the bottom of the stack,
>    and a page size can be as small as 4096 bytes.  So we cannot safely
>    allocate anything larger than 4096 bytes.  Also care for the possibility
>Index: regexec.c
>===================================================================
>RCS file: /sources/gawk/gawk-stable/regexec.c,v
>retrieving revision 1.5
>diff -u -r1.5 regexec.c
>--- regexec.c   2 Feb 2007 08:47:18 -0000       1.5
>+++ regexec.c   8 Feb 2007 21:53:02 -0000
>@@ -1411,9 +1411,11 @@
>   cur_node = dfa->init_node;
>   re_node_set_init_empty (&eps_via_nodes);
>
>+#ifdef HAVE_ALLOCA
>   if (__libc_use_alloca (nmatch * sizeof (regmatch_t)))
>     prev_idx_match = (regmatch_t *) alloca (nmatch * sizeof (regmatch_t));
>   else
>+#endif
>     {
>       prev_idx_match = re_malloc (regmatch_t, nmatch);
>       if (prev_idx_match == NULL)
>@@ -2892,7 +2894,7 @@
>              sizeof (re_dfastate_t *) * (path->alloc - old_alloc));
>     }
>
>-  str_idx = path->next_idx ?: top_str;
>+  str_idx = path->next_idx ? path->next_idx : top_str;
>
>   /* Temporary modify MCTX.  */
>   backup_state_log = mctx->state_log;
>@@ -3313,9 +3315,11 @@
>      from `state'.  `dests_node[i]' represents the nodes which i-th
>      destination state contains, and `dests_ch[i]' represents the
>      characters which i-th destination state accepts.  */
>+#ifdef HAVE_ALLOCA
>   if (__libc_use_alloca (sizeof (struct dests_alloc)))
>     dests_alloc = (struct dests_alloc *) alloca (sizeof (struct dests_alloc));
>   else
>+#endif
>     {
>       dests_alloc = re_malloc (struct dests_alloc, 1);
>       if (BE (dests_alloc == NULL, 0))
>
>-=-=-=-=-=-


-- 
Aharon (Arnold) Robbins --- Pioneer Consulting Ltd.     arnold AT skeeve DOT com
P.O. Box 354            Home Phone: +972  8 979-0381    Fax: +1 206 350 8765
Nof Ayalon              Cell Phone: +972 50  729-7545
D.N. Shimshon 99785     ISRAEL