diff options
Diffstat (limited to 'README_d')
| -rw-r--r-- | README_d/README.irix | 55 | ||||
| -rw-r--r-- | README_d/README.sgi | 28 | ||||
| -rw-r--r-- | README_d/README.tests | 46 |
3 files changed, 57 insertions, 72 deletions
diff --git a/README_d/README.irix b/README_d/README.irix deleted file mode 100644 index f69bf347..00000000 --- a/README_d/README.irix +++ /dev/null @@ -1,55 +0,0 @@ -From: "Kaveh R." <ghazi@caip.rutgers.edu!> -Date: Fri, 6 Dec 1996 11:51:27 -0500 (EST) -To: arnold@skeeve.atl.ga.us -Subject: Re: new (and I sure hope final) test release of 3.0.1 - - Gawk built everywhere except on irix-4.0.5: - -cc -c -DGAWK -I. -I.. -DHAVE_CONFIG_H ../array.c -accom: Error: ../protos.h, line 73: One declaration for is non-prototype style,) - extern int fprintf () ; - ----------------------^ -accom: Error: ../protos.h, line 73: One declaration for fprintf is non-prototyp) - extern int fprintf () ; - ----------------------^ -accom: Error: ../protos.h, line 78: redeclaration of fwrite - extern int fwrite(); - -------------------^ -accom: Error: ../protos.h, line 91: One declaration for is non-prototype style,) - extern int sscanf () ; - ---------------------^ -accom: Error: ../protos.h, line 91: One declaration for sscanf is non-prototype) - extern int sscanf () ; - ---------------------^ -accom: Error: ../protos.h, line 126: One declaration for is non-prototype style) - extern int sprintf () ; - -----------------------------^ -accom: Error: ../protos.h, line 126: One declaration for sprintf is non-prototy) - extern int sprintf () ; - -----------------------------^ -accom: Error: /usr/include/unistd.h, line 96: One declaration for is non-protot) - extern int execl( char *, char *, ...); - -----------------------------------------------^ -accom: Error: /usr/include/unistd.h, line 96: One declaration for execl is non-) - extern int execl( char *, char *, ...); - -----------------------------------------------^ -make: *** [array.o] Error 1 - - - The compiler on irix-4.0.5 understands and uses prototypes in -its headers, but does not define __STDC__. Since gawk's source code -prototypes various systems calls and libc functions, without function -arguments since __STDC__ is not defined, you get these problems. - - I was able to get around it by setting CFLAGS to "-D__STDC__=1". -Then gawk's source code prototyped functions with arguments, and the above -errors did not occur. I wonder why gawk is prototyping system functions. - - Anyway, after using the CFLAGS hack (only necessary on irix-4.0.5) -all systems built and passed "make check". - - --Kaveh --- -Kaveh R. Ghazi Systems Engineer / Project Leader -ghazi@caip.rutgers.edu ICon CMT Corp. - diff --git a/README_d/README.sgi b/README_d/README.sgi index 69b11cd5..baf74734 100644 --- a/README_d/README.sgi +++ b/README_d/README.sgi @@ -1,20 +1,14 @@ -From emory!hawkwind.utcs.toronto.edu!cks Mon Dec 30 20:12:35 1996 -Return-Path: <emory!hawkwind.utcs.toronto.edu!cks> -To: arnold@skeeve.atl.ga.us (Arnold D. Robbins) -Subject: Re: gawk regex bug -In-Reply-To: Your message of Tue, 30 Jul 1996 21:45:00 -0400. - <m0ulQME-000GWyC@skeeve.atl.ga.us> -Date: Mon, 30 Dec 1996 19:37:30 -0500 -From: Chris Siebenmann <emory!hawkwind.utcs.toronto.edu!cks> -Message-Id: <96Dec30.193738est.24603@hawkwind.utcs.utoronto.ca> -Status: OR -Content-Length: 196 -X-Lines: 5 -X-Display-Position: 2 +Tue May 2 11:40:54 IDT 2000 - The latest in my 'gawk on SGIs' stuff: with the newest SGI compilers -(MIPSPro 7.1, just released) gawk 3.0 compiles and passes the selftests -even at the highest (-O3) optimization level. +GCC and gawk often don't mix on SGI systems. Use the native C compiler to +compile gawk. `make test' should work ok, although the `tweakfld' test +may fail. That's ok; see README.ultrix for the details on that one. - - cks +Note that the SGI compiler will complain about some constructs in +regex.c and dfa.c. It's ok to ignore those complaints. +If you ask me about this, I will fuss at you for not having done +your homework! + +Arnold Robbins +arnold@gnu.org diff --git a/README_d/README.tests b/README_d/README.tests new file mode 100644 index 00000000..1c969d75 --- /dev/null +++ b/README_d/README.tests @@ -0,0 +1,46 @@ +Date: Sat, 22 Apr 2000 06:07:06 -0600 (MDT) +From: "Nelson H. F. Beebe" <beebe@math.utah.edu> +To: arnold@gnu.org +Cc: beebe@math.utah.edu, sysstaff@math.utah.edu, othmer@math.utah.edu +Subject: gawk-3.0.4 and a GNU/Linux gotcha + +Yesterday, I was assisting a colleague install some software on his +GNU/Linux machine for which uname -r reports 2.2.14. + +A (mis)feature of this system, which I've never encountered before, +broke the build of one of my programs, and also of gawk-3.0.4. + +Namely, the kernel will not execute anything that resides in /tmp, +though it will if the same script is in /usr/tmp! + +% cat /tmp/foo.sh +#! /bin/sh +echo hello + +ls -l /tmp/foo.sh +-rwxr-xr-x 1 othmer math 22 Apr 21 10:34 /tmp/foo.sh* + +% /tmp/foo.sh +bash: /tmp/foo.sh: Permission denied + +% cp /tmp/foo.sh /usr/tmp + +% /usr/tmp/foo.sh +hello + +Thus, programs that do a temporary install in /tmp, as some of mine do +in order to run the validation suite, will fail. + +gawk-3.0.4, and likely other gawk versions, hits this problem too. It +fails because test/poundbang starts with + +#! /tmp/gawk -f + +I tracked down where it comes from: + +% grep /tmp /etc/fstab +/dev/hda3 /tmp ext2 rw,nosuid,noexec,nouser,auto,async,nodev 1 1 + !!!!!! + +Since this is done via a mount command, potentially ANY directory tree +could be mounted with noexec. |