[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: started applying Marcus' patch

From: Marcus Comstedt <marcus_at_mc.pp.se>
Date: 2002-07-19 01:46:41 CEST

=?UTF-8?B?QnJhbmtvIMSMaWJlag==?= <brane@xbc.nu> writes:

> >should I take it that you are opposed to the very idea of
> >letting the user use properties for both binary and text data?
> >
>
> Um, they already can, even though Subversion doen't know about what
> they're using them for.

You can't have it store characters for you. Only octets. If you want
to use the properties for text, storing characters is much more
useful. Especially since there are no metaproperties in which to
declare a character encoding to interpret the octets by.

> >May I ask why?
> >
>
>
> Because, if we go down this road, then sooner or later we'll start
> recording the prop values mime type, newline convention, ...

You say it like it's a bad thing. :-)

Actually, newline conventions _is_ something that might need
addressing. Let's say you create a svn:ignore containing multiple
lines on UNIX (using LF newline conversion) and then propedit it on a
Macintosh (using CR newline conversion). Won't it come out wrong in
the Macintosh editor?

> User-defined props will presumably be application-specific. If users
> or their applications care about the distinction, they can invent
> their own naming scheme, magic header, or whatever. They can even
> validate and modify the values using a pre-commit hook.

Yes. But it seems a bit silly to require the user to go through all
that work when the functionality can be easily (and more cleanly)
provided by the system. It's a bit like saying "Oh, why do we need a
log? People can make their own by writing commit-hooks".

> A long time ago, somebody advanced the idea that we should enforce a
> property naming scheme (e.g., like the reverse-domain-name scheme used
> in Java), so that different user applications wouldn't tromp on each
> other's props. It was turned down for much the same reasons.

I don't think that is exactly the same thing, so if this is an attempt
to kill discussion before it's even started I don't buy it. :-)

> BTW, the fact that we munge file contents has no bearing on the
> issue. We know for sure that newline conversion and keyword expansion
> are commonly used, useful operations, expecially in software
> development environmnets. We have no idea how users will use props, or
> whether the code we put in now will ever be used outside our
> regression tests.

That goes for all code of course. The goal must be to add code which
we consider useful and would want to use ourselves. Then we'll have
to see how it fits other peoples requirements.

> Maybe we'll revisit this issue in 2.0, after we've had time to gather
> enough usage data (and specific feature requests).

It's hard to gather usage data for functionality that doesn't exist
though. Maybe it's exactly the lack of this functionality that would
make people not use properties. Who knows. Only thing I know for
sure is that _I_ would not want to use them without it.

  // Marcus

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Jul 19 01:52:58 2002

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.