Re: make/tests issues

[Paul D. Smith]

%% "J. Grant" <address@hidden> writes:

  jg> Hmm, do you know which win32 configurations are known to work?

You should ask on the address@hidden mailing list.  That's where the
Windows ports are discussed.

  jg> MSYS/rxvt should be fine, I have perl, diff installed.  This bug
  jg> is also present when using cmd.exe so I don't think it can be my
  jg> shell.

I don't either.

I think it's a line ending difference thing: DOS/Windows uses CRLF line
endings, while UNIX uses LF.

There's apparently some difference between what Perl generates when it
does a 'print "\n";' and what C generates when it does a 'printf("\n");'
in terms of line endings.

You'll have to ask on the make-w32 list or else a Perl list that deals
with Windows, or check the web or something; I don't have a Windows
machine much less one with any sort of development environment on it.

  jg> make/tests/work/features/comments.diff  has it has 0A line ending for
  jg> the first comparison, and 0D0A for the second.

Yes, this is CRLF.

  jg> I am not very familiar with the perl langage.  I added some debug print
  jg> code, and fixed a bug (the scripts continued after it was detected that
  jg> there was no make binary on the system previously).  I attatch the log
  jg> (script_error3.txt) of the commands leading upto the removal of
  jg> "c:\bin\make.exe".  Seems like this line is causing the error possibly:

  jg> system "add_profile $make_path";

???

You're saying this call is getting executed?  That should only happen on
VMS systems, not on Windows; you can tell because the test around it is
only true when $vos is set; $vos is only set (in theory) when Perl
detects that it's running on VMS.

Bleah, looking at it the whole contents of the get_osname() function in
test_driver.pl seems kind of broken.

  jg> I cannot spot why the code is removing "c:\bin\make.exe"
  jg> though. Any ideas?

None.

  >> 
  jg> GNU Make %VERSION%
  >> 
  >> Note that you haven't set the version.

  jg> Yes, this was my "unmodifed latest CVS", I want to ensure it all works
  jg> before I test my patch.  This %VERSION% should not make any difference
  jg> to the result though should it?  I changed it back to "JG_CVS" just in 
case.


  jg> I also attatch a log illustrating another bug with a path that
  jg> ends up converted from "c:\bin\make.exe" to
  jg> `c:binmake.exe.features_comments*' a little while before the exe
  jg> is deleted.

I think the reason for this is because the test script has decided that
your system is running VMS.

If you solve that problem I'm sure things will start working _MUCH_
better.

-- 
-------------------------------------------------------------------------------
 Paul D. Smith <address@hidden>          Find some GNU make tips at:
 http://www.gnu.org                      http://make.paulandlesley.org
 "Please remain calm...I may be mad, but I am a professional." --Mad Scientist