Greg Hudson wrote:
> On Sat, 2004-11-27 at 06:16, Max Bowsher wrote:
>> I think it is better that it errors immediately, rather than potentially
>> after spent a non-trivial time processing data.
>
> It'll only make it as far as the first file.
>
>> Any thoughts on my proposal to remove the _FORMAT_VERSION define, on the
>> grounds that it has non non-buggy use?
>
> Er, yes, my thought was that we should make it have a non-buggy use,
> rather than removing it and violating our API policies.
I'm not sure that any non-buggy use exists.
I feel that it's existance is an invitation to misuse it.
We have no API that returns a dumpstream version to user code, thus the only
use is in writing dumpstreams. Any dumpstream writing code should be writing
the version number of the format that it is actually outputting, not
whatever happens to be the highest version parseable by the subversion
version it was compiled against.
Max.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Nov 28 16:58:07 2004