You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Karl Fogel <kf...@red-bean.com> on 2008/06/02 02:04:46 UTC
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Blair Zajac <bl...@orcaware.com> writes:
> If we're concerned about people using pre 1.5 clients, then we should
> update the collab.net start-commit to check for the mergeinfo
> capability.
>
> Unfortunately, we can't check for capabilities in the pre-commit
> script, so you couldn't check that capability only for paths on
> branches/, so people could still use < 1.5 on trunk.
Two questions:
1. In the future, should we add a capabilities check in pre-commit?
(I haven't thought deeply about what's involved in doing that, but
hey, we could at least get newer clients and servers doing it
reliably -- the rest is a compatibility dance).
2. I don't quite understand your last paragraph. How is whether or
not we can check a capability related to which paths the user is
committing on? Could you say it more verbosely, maybe?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Posted by David Glasser <gl...@davidglasser.net>.
On Mon, Jun 2, 2008 at 2:35 PM, Karl Fogel <kf...@red-bean.com> wrote:
> "David Glasser" <gl...@davidglasser.net> writes:
>>> 1. In the future, should we add a capabilities check in pre-commit?
>>> (I haven't thought deeply about what's involved in doing that, but
>>> hey, we could at least get newer clients and servers doing it
>>> reliably -- the rest is a compatibility dance).
>>>
>>> 2. I don't quite understand your last paragraph. How is whether or
>>> not we can check a capability related to which paths the user is
>>> committing on? Could you say it more verbosely, maybe?
>>
>> We could pass information from start-commit to pre-commit by setting a
>> txnprop. This is almost certainly a bad idea.
>
> Well, more to the point, it wouldn't be "us" that does that -- it would
> be the hooks themselves, which hook authors are free to do right now.
> (Interhook information passing is a time-honored tradition over in
> CVS-land, by the way.)
Right, I meant "we" in the sense of "the maintainers of the
svn.collab.net repositories".
--dave
--
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Posted by Karl Fogel <kf...@red-bean.com>.
"David Glasser" <gl...@davidglasser.net> writes:
>> 1. In the future, should we add a capabilities check in pre-commit?
>> (I haven't thought deeply about what's involved in doing that, but
>> hey, we could at least get newer clients and servers doing it
>> reliably -- the rest is a compatibility dance).
>>
>> 2. I don't quite understand your last paragraph. How is whether or
>> not we can check a capability related to which paths the user is
>> committing on? Could you say it more verbosely, maybe?
>
> We could pass information from start-commit to pre-commit by setting a
> txnprop. This is almost certainly a bad idea.
Well, more to the point, it wouldn't be "us" that does that -- it would
be the hooks themselves, which hook authors are free to do right now.
(Interhook information passing is a time-honored tradition over in
CVS-land, by the way.)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Posted by ep...@google.com.
"David Glasser" <gl...@davidglasser.net> writes:
> We could pass information from start-commit to pre-commit by setting a
> txnprop. This is almost certainly a bad idea.
start-commit runs before there's a txn; it only gets repo path
and username as arguments.
+1 to adding capabilities as a pre-commit argument.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Posted by David Glasser <gl...@davidglasser.net>.
On Sun, Jun 1, 2008 at 10:04 PM, Karl Fogel <kf...@red-bean.com> wrote:
> Blair Zajac <bl...@orcaware.com> writes:
>> If we're concerned about people using pre 1.5 clients, then we should
>> update the collab.net start-commit to check for the mergeinfo
>> capability.
>>
>> Unfortunately, we can't check for capabilities in the pre-commit
>> script, so you couldn't check that capability only for paths on
>> branches/, so people could still use < 1.5 on trunk.
>
> Two questions:
>
> 1. In the future, should we add a capabilities check in pre-commit?
> (I haven't thought deeply about what's involved in doing that, but
> hey, we could at least get newer clients and servers doing it
> reliably -- the rest is a compatibility dance).
>
> 2. I don't quite understand your last paragraph. How is whether or
> not we can check a capability related to which paths the user is
> committing on? Could you say it more verbosely, maybe?
We could pass information from start-commit to pre-commit by setting a
txnprop. This is almost certainly a bad idea.
--dave
--
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r31535 - in branches/1.5.x: . subversion/po
Posted by Blair Zajac <bl...@orcaware.com>.
Karl Fogel wrote:
> Blair Zajac <bl...@orcaware.com> writes:
>> If we're concerned about people using pre 1.5 clients, then we should
>> update the collab.net start-commit to check for the mergeinfo
>> capability.
>>
>> Unfortunately, we can't check for capabilities in the pre-commit
>> script, so you couldn't check that capability only for paths on
>> branches/, so people could still use < 1.5 on trunk.
>
> Two questions:
>
> 1. In the future, should we add a capabilities check in pre-commit?
> (I haven't thought deeply about what's involved in doing that, but
> hey, we could at least get newer clients and servers doing it
> reliably -- the rest is a compatibility dance).
>
> 2. I don't quite understand your last paragraph. How is whether or
> not we can check a capability related to which paths the user is
> committing on? Could you say it more verbosely, maybe?
Well, this shouldn't be an issue for long, but we have a class of committers who
don't have 1.5. For example, my Ubuntu Dapper Drake server has svn 1.3 on it.
It would be nice if you are working on trunk and not doing merges, then you can
use any svn client to commit. However, if you're working on the 1.5.x branch,
then we should require a 1.5 client.
So you could have a pre-commit that checks the paths being commited with the txn
and then checks the capabilities if the path being committed should always have
mergeinfo.
Right now, it's an all or nothing approach with the entire repository.
This wouldn't work with reintegrate merges into trunk, but hopefully people
doing that would know to use a 1.5 client.
Blair
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org