On Sat, Oct 20, 2007 at 11:15:27AM -0700, Matthew Veenstra wrote:
> We deal with a lot of binary files in our CVS and my recommendation for cross
> platform/tool support is to use binary in a case like this.  This guarantees
> file integrity at the expense of easy diff'ing and merging/updating.  These
> files do not change much so I don't see a big drawback here.
>
> Maybe there are cases where we want curl to test each platform with text in
> native line feeds, but in those cases I think we are better off to have a test
> case for each line feed option from, Unix, Mac, to Win and put those on the CVS
> as a binary file and have curl handle all of them as needed.  
>
> I cannot see the drawback to using binary here (other than changing them on the
> CVS is a pain), but maybe I am missing something.  

You bring up a good point that I forgot--CR/LF characters *are*
significant in some of the test data files (and probably ought to be
significant in more--they're not applied very consistently). If the
file format iis ever converted to pure XML (with character entity support)
this won't be the case any more, but for now, I'm going to change my
mind and agree that all binary is the better way to go.

Also note that some sections support mode="text" which tells the test suite
to ignore any end-of-line differences for certain test sections. The
presence of that attribute implies a binary CVS file format.

>>> Dan

-- 
http://www.MoveAnnouncer.com              The web change of address service
          Let webmasters know that your web site has moved