sed-devel
[Top][All Lists]
Advanced

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

Re: sed-4.7.13-3c1e on mingw


From: Jim Meyering
Subject: Re: sed-4.7.13-3c1e on mingw
Date: Sat, 11 Jan 2020 19:19:01 -0800

On Mon, Jan 6, 2020 at 5:07 AM Bruno Haible <address@hidden> wrote:
>
> On mingw (from within a Cygwin build environment), the compilation
> succeeds, but there are 45 test failures. Log attached.
>
> 1) Many of the failures are obviously due to differences in end-of-line
> convention, between the expected output and the actual output.
> In the GNU gettext tests, I use this idiom to ignore these differences:
>
> LC_ALL=C tr -d '\r' < output.tmp > output || exit 1
>
> 2) Other test failures occur because mingw shows the absolute program name,
> whereas other platforms don't:
>   sed: -e expression #1, char 0: no previous regular expression
> versus
>   C:\cygwin64\home\bruno\sed-4.7.13-3c1e\build-mingw64\sed\sed.exe: -e 
> expression #1, char 0: no previous regular expression
> In the GNU gettext tests, I postprocess the standard error like this:
>
> _some_command_ 2>&1 | grep -v '^==' | sed -e 's|[^ ]*\\sed\.exe|sed|' -e 
> 's|^sed\.exe|sed|' -e 's|^/cygdrive/[^ ]*/sed|sed|' | LC_ALL=C tr -d '\r' > 
> out
>
> The "grep -v '^=='" removes valgrind output. The next pipe normalizes
> native Windows and Cygwin program names. The final 'tr' command converts
> to Unix end-of-line conventions.

Nice. Thanks for all the suggestions. However, taking those
suggestions would be very invasive to all tests, and making the tests
work on CR/NL line terminating systems is pretty low priority for me,
so I'm inclined to put this off until after the release, in case
someone contributes patches.



reply via email to

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