make-w32
[Top][All Lists]
Advanced

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

Re: Schedule for GNU make 3.81


From: Alessandro Vesely
Subject: Re: Schedule for GNU make 3.81
Date: Thu, 24 Feb 2005 12:21:54 +0100

"Paul D. Smith" wrote:
> 
> %% "J. Grant" <address@hidden> writes:
> 
>   jg> This sounds good, it would save me having to put a dos2unix
>   jg> command or use special diff options.
> 
>   jg> Previously I tried to use a DJ diff.exe, but that did not work
>   jg> with the rest of the testing process (if I remember correctly).
>   jg> Using dos2unix.exe did work, but then unlink was deleting my MSVC
>   jg> win32 make.exe instead of a temp file.

I had the same problem with the executable being deleted, but I don't think
it is anyhow related to line terminations.
 
> It seems like the simplest thing to do, from a portability standpoint,
> would be for the regression test Perl script to fix up the output file
> itself, before invoking diff on it.

This must deped on the perl port. With ActiveState Perl (for win32) I got
both base and log files correctly terminated with \r\n. Thus, no problems
with diff on that.

> It seems like it should be simple enough to rewrite the
> run_command_with_output() sub in test_driver.pl to first capture the
> output rather than redirecting it to the file, then translations can be
> made to the output, then it could be written to the file.

Otherwise, both base and log could be translated just before (or after)
running diff. Another quirk that translation can solve is backslashes:

 D:\tmp\TOOLS\make\make-3.81beta2\tests\work\features>head -2 recursion.log
 C:\UTIL\GMAKE.EXE: Entering directory `D:/tmp/TOOLS/make/make-3.81beta2/tests'
 C:/UTIL/GMAKE.EXE -f work/features/recursion.mk foo

 D:\tmp\TOOLS\make\make-3.81beta2\tests\work\features>head -2 recursion.base
 C:\UTIL\GMAKE.EXE: Entering directory `D:\tmp\TOOLS\make\make-3.81beta2\tests'
 C:/UTIL/GMAKE.EXE -f work/features/recursion.mk foo

I wouldn't consider this an error; it might be signalled as a warning (in the
sense that it is probably an error of the test script.) After running diff,
one can translate both files and, if that fixes it, count a warning.




reply via email to

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