You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apr.apache.org by rb...@covalent.net on 2001/02/11 16:26:47 UTC
Poliicy proposal.
One of the things that I have been doing a lot recently, is tagging the
tree with APACHE_2_0_* tags. This is because Apache 2.0 is trying to hit
beta, and we are rolling a lot of trees. Apache 2.0 needs our builds to
be reproducable, and I am sure every other project does too. I figure
there are two sets of people who will use APR in their projects.
1) Projects that have somebody with commit access in APR, will most
likely use HEAD, just like Subversion and Apache do.
2) Other projects will likely use a specific tag or tarball.
I would like to suggest that ANY project with a committer on the list can
tag the tree at any time. Those tags should be in a format like
APACHE_2_0 or SUBVERSION_X_Y. When APR releases a tarball, we will use
APR_X_Y, which will allow people to easily distinguish why the tag was
added.
If this is adopted, I'll write a small web page about it, so the site is
updated.
Ryan
_______________________________________________________________________________
Ryan Bloom rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------
Re: Poliicy proposal.
Posted by Greg Stein <gs...@lyra.org>.
Oh... summary of what I said: Ryan's approach is the right one for now. If
somebody wants to spend some time formalizing APR's release process, then go
for it.
Cheers,
-g
On Mon, Feb 12, 2001 at 01:58:11PM -0800, Greg Stein wrote:
> On Mon, Feb 12, 2001 at 01:04:16PM -0800, Daniel Rall wrote:
> > Ben Hyde <bh...@pobox.com> writes:
> > > rbb@covalent.net wrote:
> > > > On Mon, 12 Feb 2001, Ben Hyde wrote:
> > > > > > APACHE_2_0 or SUBVERSION_X_Y.
> > > > >
> > > > > I find this weird
> > >
> > > still
> > [...]
> > > > as easy as possible.
> > >
> > > it's weird for tags in apr to named after user's projects.
> > [...]
> > > meanwhile the apr using code should be paranoid
> > > at compile & runtime about what version of apr
> > > they are associating with.
> > >
> > > sequential fast moving build numbers are good
> > > for this.
> >
> > I agree with Ben. What's the problem with doing another tag of APR
> > whenever one of the two projects driving development needs a tag for
> > its own build?
>
> APR does not (yet) have its own independent release system. It is not yet
> formally released with its own version numbers and labelling.
>
> As a further result, none of the clients are including APR based on a
> tarball. They all refer directly to the HEAD and release the APR sources
> bundled directly in.
>
> Any tags that the clients are placing on the APR tree are merely commentary.
> Something along the lines of "we snapped a release right <here>".
>
> If somebody wants to take a bit of time and turn the APR release mechanics
> into a formal system, then we can start adjusting the Apache and SVN release
> systems to say "we chose APR 1.0.7 for this release of <FOO>".
>
> Cheers,
> -g
>
> --
> Greg Stein, http://www.lyra.org/
--
Greg Stein, http://www.lyra.org/
Re: Poliicy proposal.
Posted by Greg Stein <gs...@lyra.org>.
On Mon, Feb 12, 2001 at 01:04:16PM -0800, Daniel Rall wrote:
> Ben Hyde <bh...@pobox.com> writes:
> > rbb@covalent.net wrote:
> > > On Mon, 12 Feb 2001, Ben Hyde wrote:
> > > > > APACHE_2_0 or SUBVERSION_X_Y.
> > > >
> > > > I find this weird
> >
> > still
> [...]
> > > as easy as possible.
> >
> > it's weird for tags in apr to named after user's projects.
> [...]
> > meanwhile the apr using code should be paranoid
> > at compile & runtime about what version of apr
> > they are associating with.
> >
> > sequential fast moving build numbers are good
> > for this.
>
> I agree with Ben. What's the problem with doing another tag of APR
> whenever one of the two projects driving development needs a tag for
> its own build?
APR does not (yet) have its own independent release system. It is not yet
formally released with its own version numbers and labelling.
As a further result, none of the clients are including APR based on a
tarball. They all refer directly to the HEAD and release the APR sources
bundled directly in.
Any tags that the clients are placing on the APR tree are merely commentary.
Something along the lines of "we snapped a release right <here>".
If somebody wants to take a bit of time and turn the APR release mechanics
into a formal system, then we can start adjusting the Apache and SVN release
systems to say "we chose APR 1.0.7 for this release of <FOO>".
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
Re: Poliicy proposal.
Posted by Daniel Rall <dl...@collab.net>.
Ben Hyde <bh...@pobox.com> writes:
> rbb@covalent.net wrote:
> > On Mon, 12 Feb 2001, Ben Hyde wrote:
> > > > APACHE_2_0 or SUBVERSION_X_Y.
> > >
> > > I find this weird
>
> still
[...]
> > as easy as possible.
>
> it's weird for tags in apr to named after user's projects.
[...]
> meanwhile the apr using code should be paranoid
> at compile & runtime about what version of apr
> they are associating with.
>
> sequential fast moving build numbers are good
> for this.
I agree with Ben. What's the problem with doing another tag of APR
whenever one of the two projects driving development needs a tag for
its own build?
--
Daniel Rall <dl...@finemaltcoding.com>
Re: Poliicy proposal.
Posted by Ben Hyde <bh...@pobox.com>.
rbb@covalent.net wrote:
> On Mon, 12 Feb 2001, Ben Hyde wrote:
> > > APACHE_2_0 or SUBVERSION_X_Y.
> >
> > I find this weird
still
> The problem I am trying to solve,
hear hear
> as easy as possible.
it's weird for tags in apr to named after user's projects.
> kept a copy of the tree in their repositories.
or the symbolic equivalent of that.
if a customer needs apr to do (what ever you want to call
it, freeze, release, tag, varnish) then apr should.
meanwhile the apr using code should be paranoid
at compile & runtime about what version of apr
they are associating with.
sequential fast moving build numbers are good
for this.
- ben
Re: Poliicy proposal.
Posted by rb...@covalent.net.
On Mon, 12 Feb 2001, Ben Hyde wrote:
>
> > Those tags should be in a format like
> > APACHE_2_0 or SUBVERSION_X_Y.
>
> I find this weird. Library providers are not supposed to know such
> things about thier clients. It's the job the clients to record what
> version of the library they are dependent upon. Surely this tag is just
> an attempt to defer the day when the apr clients have to do this? Think
> MODULE_MAGIC_NUMBER...
The problem I am trying to solve, is that there are currently two projects
that are driving development of APR, which means that both are using HEAD
of the APR tree. Neither project can really use another tag, because they
are driving the development, and thus need to be on the cutting
edge. When one of those projects wants to make a release, they need to be
able to reproduce that release easily. I believe 99% of the apr clients
will use a simple tag from the APR tree, but those that must be on the
cutting edge will always find that hard to do. I am trying to make this
as easy as possible.
That's my goal anyway. I expect that when APR actually makes a beta or
release version, then a good portion of this problem will go away. This
would also be easier if Apache and Subversion didn't ask the developers to
check out APR, but instead kept a copy of the tree in their repositories.
Ryan
_______________________________________________________________________________
Ryan Bloom rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------
Re: Poliicy proposal.
Posted by Ben Hyde <bh...@pobox.com>.
> Those tags should be in a format like
> APACHE_2_0 or SUBVERSION_X_Y.
I find this weird. Library providers are not supposed to know such
things about thier clients. It's the job the clients to record what
version of the library they are dependent upon. Surely this tag is just
an attempt to defer the day when the apr clients have to do this? Think
MODULE_MAGIC_NUMBER...
- ben
Re: Poliicy proposal.
Posted by rb...@covalent.net.
On 12 Feb 2001, Karl Fogel wrote:
> Jeff Trawick <tr...@bellsouth.net> writes:
> > > I would like to suggest that ANY project with a committer on the list can
> > > tag the tree at any time. Those tags should be in a format like
> > > APACHE_2_0 or SUBVERSION_X_Y. When APR releases a tarball, we will use
> > > APR_X_Y, which will allow people to easily distinguish why the tag was
> > > added.
> >
> > This sounds very useful to me.
>
> +1
That's either three or four (I didn't really count), so I'll update the
web page with this information.
Ryan
_______________________________________________________________________________
Ryan Bloom rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------
Re: Poliicy proposal.
Posted by Karl Fogel <kf...@galois.collab.net>.
Jeff Trawick <tr...@bellsouth.net> writes:
> > I would like to suggest that ANY project with a committer on the list can
> > tag the tree at any time. Those tags should be in a format like
> > APACHE_2_0 or SUBVERSION_X_Y. When APR releases a tarball, we will use
> > APR_X_Y, which will allow people to easily distinguish why the tag was
> > added.
>
> This sounds very useful to me.
+1
Re: Poliicy proposal.
Posted by Jeff Trawick <tr...@bellsouth.net>.
rbb@covalent.net writes:
> I would like to suggest that ANY project with a committer on the list can
> tag the tree at any time. Those tags should be in a format like
> APACHE_2_0 or SUBVERSION_X_Y. When APR releases a tarball, we will use
> APR_X_Y, which will allow people to easily distinguish why the tag was
> added.
This sounds very useful to me.
--
Jeff Trawick | trawickj@bellsouth.net | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married an alien...
Re: Poliicy proposal.
Posted by cm...@collab.net.
rbb@covalent.net writes:
> I would like to suggest that ANY project with a committer on the list can
> tag the tree at any time. Those tags should be in a format like
> APACHE_2_0 or SUBVERSION_X_Y. When APR releases a tarball, we will use
> APR_X_Y, which will allow people to easily distinguish why the tag was
> added.
+1.
But don't expect to see SUBVERSION_X_Y tags for a while... :-)