Julian Foad wrote:
> Branko Čibej wrote:
>
>> kfogel@tigris.org wrote:
>>
>>> + if (svn_ctype_iscntrl (*q)
>>> + && ! ((*q == '\n') || (*q == '\r') || (*q == '\t')))
>>>
>> These are UTF-8 strings, so you should define character value
>> constants in svn_ctype.h and use them here.
>
>
> Are you sure that we should? I believe we use "\n" etc. all over the
> place in UTF-8 strings. It would only be wrong in a
> non-ASCII-compatible system, and the only one that I have heard of
> Subversion being ported to is an EBCDIC one that was discussed on the
> list and presumably has a satisfactory way of handling this.
We define constants for other ASCII codes. We should be consistent and
define these, too. AFAIK that EBCDIC port uses exactly this kind of
"satisfactory" solution.
Besides, using an svn_ctype_* function, but then testing against
character literals, is semantically broken. We're looking at UTF-8, not
the execution charset, and we shouldn't mix the two.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Feb 10 05:26:34 2005