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

Re: [PATCH] check name svn special properties

From: Julian Foad <julianfoad_at_btopenworld.com>
Date: 2005-08-01 22:09:25 CEST

John Peacock wrote:
> Dale Worley wrote:
>
>> This sounds like a very good idea to me also. "svn:*" properties are
>> reserved to Subversion; there's no reason for anyone to create one that
>> Subversion doesn't recognize, and 99% of the time it is a mistake.
>
> Yes, but my objection to locating this code _in the client_ is still
> there. Say we add this to 1.3 and then for 1.4, a new svn: property
> gets added (say svn:log-template). Now our compatibility guidelines
> require that a 1.3 client must be able to communicate with a 1.4 server,
> but it doesn't have to get all of the new functionality. In this case,
> any 1.3 client must use --force to set the svn:log-template property.

Yes. You think that isn't generous enough? I think it is.

> So, although it is possible to work around it with the backrev'd client,
> but it might be considered a hindrance to the adoption of new
> properties. It's not likely that the project will be adding new
> properties willy-nilly, but if more are added, it would become an
> increasing nuisance to know which properties need --force and which do
> not. I know, users should keep their clients up to date, but there is
> no way to require it.

There's no way to require new clients if the users don't want the new features.
  I don't understand why you think the need to set a property that the client
doesn't understand will be more than extremely rare, or why you think the
inconvenience of specifying "--force" outweighs the advantage of error
checking. Maybe we'll just have to agree to disagree; I value your input, but
I really want to understand your perspective.

> Just my 2 cents. I'm -0.5 on the feature in the client; it should
> really be server mediated.

Most of the properties in question are entirely client-side features - at least
the node props, svn:ignore, svn:mime-type, svn:log-template, svn:eol-style,
svn:externals. It doesn't seem to make any sense for the server to mediate
their use. No? Perhaps the rev props (svn:date, svn:author, svn:log) are
mainly server-side.

Would you accept this proposal for client-side checking if we apply it only to
node props, not to rev props? Was that the main problem with it?

- Julian

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Aug 1 22:10:07 2005

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.