[Apache-Test patch] for shebang lines over 64 characters long
Stas Bekman 19 December 2005 22:30:16
Apache-Test has a problem on OpenBSD when perl's shebang is too long. This patch uses the old eval trick to workaround the problem when we see a perl path > 62 chars.
The only reason I didn't commit it right away is because I'm not sure whether this workaround works on win32. Can someone on win32 confirm that?
I'd suggest to change the patch to not check the length of $Config{perlpath}, just so that you certainly test the right thing.
Thank you.
-------- Original Message -------- Subject: Here's the patch for shebang lines over 64 characters long Date: Thu, 15 Dec 2005 14:24:22 -0800 From: Mike Smith <mike@mailchannels.com> To: Stas Bekman <stas@stason.org>
Stas Bekman 16 December 2005 01:38:53 [ permanent link ]
Stas Bekman wrote:> Apache-Test has a problem on OpenBSD when perl's shebang is too long. > This patch uses the old eval trick to workaround the problem when we see > a perl path > 62 chars.>
The only reason I didn't commit it right away is because I'm not sure > whether this workaround works on win32. Can someone on win32 confirm that?>
I'd suggest to change the patch to not check the length of > $Config{perlpath}, just so that you certainly test the right thing.
Oops, I forgot the most important part: Mike Smith <mike@mailchannels.com> wrote the patch Sorry, Mike.
Randy Kobes 18 December 2005 01:38:46 [ permanent link ]
On Fri, 16 Dec 2005, Stas Bekman wrote:
Stas Bekman wrote:>
I guess if there are no objections I'll commit it on Monday.
Sounds good to me - I tested the patch out on Win32 with the Apache-Test tests and also the apreq tests (which involve some cgi tests), and these worked fine.
Stas Bekman 18 December 2005 19:02:44 [ permanent link ]
Randy Kobes wrote:> On Fri, 16 Dec 2005, Stas Bekman wrote:>
Stas Bekman wrote:>>
I guess if there are no objections I'll commit it on Monday.>
Sounds good to me - I tested the patch out on Win32 with the> Apache-Test tests and also the apreq tests (which involve> some cgi tests), and these worked fine.
Thanks Randy, but did you try removing the < 62 check? It's only different from the current implementation if that exec trick is used (i.e. if you have a very long path to perl.exe). So try to s/62/2/ or something like that. Thanks!
Randy Kobes 18 December 2005 22:27:38 [ permanent link ]
On Sun, 18 Dec 2005, Stas Bekman wrote:
Randy Kobes wrote:>> On Fri, 16 Dec 2005, Stas Bekman wrote:>>
Stas Bekman wrote:>>>
I guess if there are no objections I'll commit it on Monday.>>
Sounds good to me - I tested the patch out on Win32 with the>> Apache-Test tests and also the apreq tests (which involve>> some cgi tests), and these worked fine.>
Thanks Randy, but did you try removing the < 62 check? It's only different > from the current implementation if that exec trick is used (i.e. if you have > a very long path to perl.exe). So try to s/62/2/ or something like that. > Thanks!
Thanks for pointing that out, Stas - I just did a s/62/2/ in the check, to force the exec trick, and both the Apache-Test and apreq tests still all passed.
Stas Bekman 19 December 2005 22:30:16 [ permanent link ]
Randy Kobes wrote:> On Sun, 18 Dec 2005, Stas Bekman wrote:>
Randy Kobes wrote:>>
On Fri, 16 Dec 2005, Stas Bekman wrote:>>>
Stas Bekman wrote:>>>>
I guess if there are no objections I'll commit it on Monday.>>>
Sounds good to me - I tested the patch out on Win32 with the>>> Apache-Test tests and also the apreq tests (which involve>>> some cgi tests), and these worked fine.>>
Thanks Randy, but did you try removing the < 62 check? It's only >> different from the current implementation if that exec trick is used >> (i.e. if you have a very long path to perl.exe). So try to s/62/2/ or >> something like that. Thanks!>
Thanks for pointing that out, Stas - I just did a s/62/2/ in> the check, to force the exec trick, and both the Apache-Test> and apreq tests still all passed.