John Peacock <jpeacock@rowman.com> writes:
> Philip Martin wrote:
>
>> I'm not sure what to do with that information :) Is it the answer to
>> one of my questions above? Please bear in mind I don't know much
>> about Windows, or OS/X, internals; in particular I don't know what
>> rules they use when determining whether two filenames match.
>
> Both OS/X and Win32 use the current locale information when performing
> case-folding, as does Perl when 'use locale;' is in force.
If OS/X and Win32 use the clients' locales to do filesystem
case-folding then your server script can only work if it uses the same
locale. Which means that it might work in the case where all the
clients use the same locale, but it won't work in general.
If the clients and server are using different OS's (which is after all
the reason for using this script) then you have also got to worry
about whether Linux/BSD/Windows/OSX all use the same rules when
implementing LC_COLLATE.
> The "problem" is that Subversion keeps track of the filenames in a
> private text file and the file system doesn't care. It is Subversion
> which notes that the file already exists when adding the second
> version; if this could be made a warning in Subversion, the file
> system would have no problem making do (literally copying the contents
> of the second file on top of the first file).
I don't see how that could be considered "making do". I suppose
checkout could be made to do what you suggest, but if the text-base
for the second file overwrites the text-base for the first file any
subsequent attempt to update the first file will fail (either with a
checksum error or because applying the first file's svndiff to second
file's text-base produces rubbish). The result is a working copy that
it is not safe to update.
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jan 31 20:44:21 2005