On Nov 24, 2004, at 9:47 AM, C. Michael Pilato wrote:
> Ben Collins-Sussman <sussman@collab.net> writes:
>
>>> Since Subversion shouldn't limit the OS capabilities arbitrarily,
>>
>> ...why not?
>
> Because there's no compelling reason not to support it?
Sure there is. It adds complexity, and I maintain that it's a feature
that no linux user cares about.
> Because if we
> were in the business of limiting OS capabilities, we'd be catering to
> the lowest common denominator of support across across all known
> supported OSes, which means we'd be trying to case-protect stuff on
> Linux for the sake of Windows users, dropping symlink support,
> dropping svn:executable support, etc?
The difference here is features like svn:executable and versioned
symlinks are things that linux users have been screaming for. I've
never, in my entire life, heard a linux user get annoyed that win32
doesn't support control characters in filenames.
I think our features should cater to specific user demands; there's no
reason to bend over backwards to support essentially unknown OS
features.
> But if the only concern you have is that we add a little
> complexity (which, by the way, would be entirely contained to the
> existing code at the fringes of XML transport), that's just the cost
> of almost any feature.
>
Heh, you guys know (better than I do) exactly how complicated this
change would be. Perhaps it's just a teeny-tiny amount of effort and
complexity, with no measurable impact on performance. But then again,
I think the result is a just teeny-tiny feature than nobody will
notice. So it's essentially impossible to weight pros vs. cons here,
since both sides require a microscope to view. :-)
So I withdraw my objection. I just wonder if there aren't bigger fish
to fry.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Nov 24 17:38:58 2004