[Svnmerge] svnmerge.py init crashes (on bad characters in svn xml log)

[Jean-Philippe Daigle]

On 2009-10-07, at 12:45 PM, Raman Gupta wrote:

>> <Jean-Philippe.Daigle at solacesystems.com> wrote:
>> No, I just upgraded to 1.6.5.x yesterday before posting this, and I'm
>> getting the same error, so it's actually printing the 0x93 / 0x94  
>> characters
>> in the XML.
>
> It certainly seems like it might be a bug in svn. FYI, in UTF-8, those
> characters should be three bytes each: 0xe2809c and 0xe2809d.
>
> http://www.eki.ee/letter/chardata.cgi?ucode=201c
> http://www.eki.ee/letter/chardata.cgi?ucode=201d
>
> If you can reproduce this in a simple test case, I'm sure it would be
> useful for the svn devs.
>
> For your immediate purposes however, why not just "fix" the log
> message using:
>
> svn propset svn:log --revprop -r REV ...
>
> (make sure to configure your repo to accept that change using the
> pre-revprop-change hook: http://svnbook.red-bean.com/en/1.1/re23.html)
>
> Cheers,
> Raman

Thanks Raman! I had no idea you could rewrite a commit log message,  
but if it's stored as a revisioned property on the directory then that  
makes sense.

Doesn't look like I can fix my immediate problem, though, seeing how  
svn now complains about the data in there being invalid (and it's  
correct, that is invalid UTF-8, AFAIK). I bet making that commit  
message today would be impossible, but I just realized that SVN commit  
is so old it was ported over with cvs2svn years ago when we switched  
over to SVN.

$ svn propedit --revprop -r 1863 svn:log
svn: Valid UTF-8 data
(hex: 62 6c 69 73 68 65 72 73 2c 20 69 66 20 74 68 65 79 20 63 68 65  
63 6b 20)
followed by invalid UTF-8 sequence
(hex: 93 53 65 6e)