Re: [xml] thanks a lot ---some more help ---



Hi,

On Wed, 2006-01-18 at 10:32 -0500, Rob Richards wrote:
Daniel Veillard wrote:
On Wed, Jan 18, 2006 at 05:49:38AM -0500, Rob Richards wrote:
  
Kasimier Buchcik wrote:
    
There's no "make tests" equivalent for the win side. If you look at
the makefile for win, you'll see only two test-sections defined there.
 
      
What exactly is the difference between the "make tests" and the 
runtest/runsuite?
I had started on building out the tests in the win makefile to make 
things a bit easier to work with but if the runtest.exe and runsuite.exe 
are how the tests should be run under windows I wont bother.
    

  Seen README.tests ?
  
All that explains is what the three programs do and how to run them. 
Just wanted to know if one way was better (or more accurate) than the other.

Reason I ask is I ran the tests both ways on linux then tried it on 
Windows. Windows failed for all the ent11 and ns7 tests using runtests 
(while the nmake tests were passing as it ended up ignoring the line 
endings) so I was tracking down why. Seems grabbing those files 
result\.. (and it is only those 2 sets of files) from CVS under windows 
has windows line endings while the same files retrieved under linux has 
unix linefeeds. Its not a CVS setting as the other result files are 
retrieved correctly with unix line endings. cvs diff shows no difference 
between the files, so wondering where the linefeeds are coming from.

If you use wincvs then there's the option "Check out text files with the
Unix LF"; I guess it's unchecked on your/my side. Same for me here,
and I'm experiencing the same errors with "ent11" and "ns7". 

This is an CVS binary/text file issue. "ent11" and "ns7" are text files
in CVS, the other test files are binary, we need to change them to
binary so that linebreaks won't be changed.

Regards,

Kasimier



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