bug-gawk
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [bug-gawk] stat issue when building 32bit on 64bit machine


From: Aharon Robbins
Subject: Re: [bug-gawk] stat issue when building 32bit on 64bit machine
Date: Wed, 13 Mar 2013 22:28:43 +0200
User-agent: Heirloom mailx 12.5 6/20/10

Hello.

Re the below. Thanks for the note.

First, the current released version is 4.0.2, so you should get that
source, or use git to clone the gawk repo and then check out the
gawk-4.0-stable branch (that would be best).

I am willing to review patches, but I wonder about the portability
of using stat64 directly, and my gut feeling is that it's a bad idea.
My understanding is that instead, various compiler options should
be used to get the headers to define things properly.  This may require
autoconf changes.

If you have specific questions, please send them. Otherwise, I have
no additional advice to offer.

Thanks,

Arnold

> From: Adam Edgar <address@hidden>
> Date: Tue, 12 Mar 2013 12:50:39 -0400
> To: address@hidden
> Subject: [bug-gawk] stat issue when building 32bit on 64bit machine
>
> I am using the opencsw build of gawk 4.0.0 on an x86-64 machine. It is
> built as a 32bit binary (ie not using -m64). Everything seems to run
> fine until the system has been up long enough for the inode for FIFO's
> has exceeded the 32bit limit. You can check this on a solaris machine
> by running :
>
> echo | ls -i /proc/self/fd/0
>
> I believe the easiest solution would be to patch the upstream source to
> use stat64 instead of just stat. I've downloaded the 4.0.0 source and ran
> into some issues trying this. Particularly with some of the yacc generated
> code. I'm willing to help with the patch and to test with a system where
> 64bit inodes exist. I just believe the long term fix belongs upstream.
>
> ASE



reply via email to

[Prev in Thread] Current Thread [Next in Thread]