On Aug 19, 2005, at 12:48 PM, Philip Martin wrote:
> Well, I haven't done much Subversion coding recently, but working on
> r15828 leads me to make the following observations:
>
> 1. svn_string_compare
>
> What a !#*%!# awful name! Does it behave like strcmp or is it one
> of those new-fangled functions that do the opposite? I had to
> resort to reading the documentation. Bleah!
New-fangled? It's as old as Subversion itself! IIRC, svn_string.h
was the very first thing we wrote when we started coding back in
summer of 2000. :-)
>
> 3. Property prej files.
>
> I think it is possible for there to be mutiple property conflicts,
> and it looks as if the bits of text written to the prej file are
> not \n separated. Should we add \n to the string literals in
> svn_wc__merge_props or should append_prop_conflict do it?
>
I think that's a very old bug... it may actually still be filed in
the issuetracker! I think append_prop_conflict() should do it.
> 4. Property encoding
>
> append_prop_conflict assumes that the property values are valid
> UTF8. It's possible to have properties that are not UTF8 and
> append_prop_conflict will fail, and so the merge will abort.
> Perhaps we should escape non-UTF8 values, or simply write them
> without conversion?
Hm, perhaps we should test to see if the value is UTF8. If not, we
should probably just substitute in the phrases "a certain value", "a
different value". I don't like the idea of trying to print
potentially binary data.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 19 23:51:48 2005