[cvsnt] Re: Test script questions

[Glen Starrett]

Tony Hoyle wrote:
> Glen Starrett wrote:
> 
>> The odd line endings in the *info test output didn't match up to 
>> info_test_output.txt, but almost matched the .w32 version of the file. 
>> I altered testcvs.py and removed the " from info_test_output.32 
>> (tricky with those line endings).
> 
> 
> Which line endings??  they're just standard text files.

When I check out info_test_output.w32 tagged CVSNT_2_5_01_1927, all the 
line that are from the commitinfo script have an extra 0D at the end, so 
I see (hex in place of CR and LF):

commitinfo - testcvs /repos/testcvs 0D 0D 0A
commitinfo - all /repos/testcvs 0D 0D 0A
Checking in diff_test.txt; 0D 0A
/repos/testcvs/diff_test.txt,v  <--  diff_test.txt 0D 0A
... etc ...

The info_test_output.txt file has normal line endings, so that would be 
consistent with the way you had the script set up when I checked it out. 
  It doesn't compare to the output of the commitinfo test though -- that 
has line endings like I described in the .w32 file, without the 
quotation marks.  I believe that implies that the output from the 
commitinfo script has a flaw, but I don't know why you aren't seeing it.


> 
>> -- Should the "" have been removed from info_test_output.w32, or is 
>> that a bug?  I presume not a bug since you had altered the script to 
>> look to the other file.
> 
> 
> Win32 generates the quotes (cmd.exe parses output differently to the 
> unix shell, so the output differs slightly), so they're needed.

It doesn't appear in the *info output on my machine, and you had the 
script pointing to the unquoted, normal line ending info_test_output.txt 
before I altered it, so looks like you know that already.

> 
>> -- Are the "funny" line endings the wrong result (and therefore 
>> shouldn't be needed in info_test_output.w32)?  I would think it would 
>> be since it's 0D 0D 0A, not just 0A like it would be if it was a 
>> normal unix format.
> 
> 
> The testcvs is run as part of the release process even for test 
> revisions, and I don't release until it passes.  You shouldn't need to 
> change anything...  maybe a different python version?
> 
> I'm using ActivePython 2.3.2 build 232 (Nov 13 2003).

I'm using the same version of python, including the date and build 
(python -v confirmed it for me).

Also, it works flawlessly "as is" on my Linux box.

-- 
Glen Starrett