RE: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 04:26:22PM -0700, Justin Erenkrantz wrote: aaa rewrite belongs in: 2.0: rbb, brianp, dreid, gstein, jim, rederpj, striker, trawick, ianh, gs, bnicholes 2.1: dpejesh, chris, aaron, hb If someone would like to do a release before I check in the aaa changes, I'd be game to hold off until Monday or Tuesday. (I don't have the time to be an RM right now.) -- justin I would actually like to see a 2.0.41 release. Subversion is seeing problems with working properly against 2.0.40 because of the apr_errno.h changes. If we get a 2.0.41 out there, then we can tell people at least 2.0.41 rather than forcing them to use CVS or a CVS snapshot. I'll try to get the release done over the weekend. Hell... at least a tag. Given a tag, then you can do whatever the heck you want. Cheers, -g +1 Bill
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 04:26:22PM -0700, Justin Erenkrantz wrote: aaa rewrite belongs in: 2.0: rbb, brianp, dreid, gstein, jim, rederpj, striker, trawick, ianh, gs, bnicholes 2.1: dpejesh, chris, aaron, hb If someone would like to do a release before I check in the aaa changes, I'd be game to hold off until Monday or Tuesday. How long do you think it will take to get the auth changes stabilized enough to make a GA-quality release? If this is more than about a week, I *strongly* urge you to branch. The worst thing that could happen would be to have us run into a security problem that needs to be fixed right away. Without a branch we would have to revert all the auth changes back to a working state (you're going to make a tag anyway, right??), make the release, and then stick it all back in again. That's much uglier than just doing a branch in the first place. Also, there's no reason that the auth branch would need to keep up with changes in HEAD. Instead, those who want to work on this stuff can make it work with 2.0.41-dev, and then once it's done take a day or so to merge those changes back into HEAD. This seems much less painful to me than forcing everyone else to deal with the instability. To put it more clearly: I am -1 for any long-term breakage of the currently-stable 2.0 tree. -aaron
Re: [TALLY] aaa rewrite
On Thu, 5 Sep 2002, Aaron Bannert wrote: On Thu, Sep 05, 2002 at 04:26:22PM -0700, Justin Erenkrantz wrote: aaa rewrite belongs in: 2.0: rbb, brianp, dreid, gstein, jim, rederpj, striker, trawick, ianh, gs, bnicholes 2.1: dpejesh, chris, aaron, hb If someone would like to do a release before I check in the aaa changes, I'd be game to hold off until Monday or Tuesday. How long do you think it will take to get the auth changes stabilized enough to make a GA-quality release? If this is more than about a week, I *strongly* urge you to branch. The argument would be that they would never be stabilized until they are in HEAD since few people will use them until then. The worst thing that could happen would be to have us run into a security problem that needs to be fixed right away. Without a branch we would have to revert all the auth changes back to a working state (you're going to make a tag anyway, right??), make the release, and then stick it all back in again. That's much uglier than just doing a branch in the first place. We can make a branch for a security release after the fact, either off the last release or off the pre-auth-change tag and then build a security-fix-only release off that branch.
Re: [TALLY] aaa rewrite
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Given this argument, might I suggest doing a release before these AAA changes are started, then at least all of the patches between 2.0.40 and now can be made GA? Also gives more time before the public start wondering where things are (if the AAA changes should take a long time). :) Chris Taylor - The guy with the PS2 WebServer Email: [EMAIL PROTECTED] - PGP: http://www.x-bb.org/chris.asc - - Original Message - From: Aaron Bannert [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, September 06, 2002 12:36 AM Subject: Re: [TALLY] aaa rewrite On Thu, Sep 05, 2002 at 04:26:22PM -0700, Justin Erenkrantz wrote: aaa rewrite belongs in: 2.0: rbb, brianp, dreid, gstein, jim, rederpj, striker, trawick, ianh, gs, bnicholes 2.1: dpejesh, chris, aaron, hb If someone would like to do a release before I check in the aaa changes, I'd be game to hold off until Monday or Tuesday. How long do you think it will take to get the auth changes stabilized enough to make a GA-quality release? If this is more than about a week, I *strongly* urge you to branch. The worst thing that could happen would be to have us run into a security problem that needs to be fixed right away. Without a branch we would have to revert all the auth changes back to a working state (you're going to make a tag anyway, right??), make the release, and then stick it all back in again. That's much uglier than just doing a branch in the first place. Also, there's no reason that the auth branch would need to keep up with changes in HEAD. Instead, those who want to work on this stuff can make it work with 2.0.41-dev, and then once it's done take a day or so to merge those changes back into HEAD. This seems much less painful to me than forcing everyone else to deal with the instability. To put it more clearly: I am -1 for any long-term breakage of the currently-stable 2.0 tree. - -aaron -BEGIN PGP SIGNATURE- Version: PGPfreeware 7.0.3 for non-commercial use http://www.pgp.com iQA/AwUBPXftpyqf8lmE2RZkEQJS/QCeKcEsBTzgYofbe8L7RL3aQjhWfNkAoJR6 QlSMRcARQO59PY0OiGjdwUlT =0gNd -END PGP SIGNATURE-
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 04:41:38PM -0700, Marc Slemko wrote: The argument would be that they would never be stabilized until they are in HEAD since few people will use them until then. And my counter argument is that it doesn't really matter if nobody uses it until it makes it into HEAD. This change, like all other changes made to httpd, should follow a couple simple rules: 1) If you're making a big change, post your patches. This means that you implement it as well as you can before posting. 2) Don't break the server, and if you do break it, make your fixes ASAP. Correct me if I'm wrong here, but it seems that the proposed changes are only theoretical, and that the features are not yet implemented. If this is not the case, and all the code is already there, then I completely drop my request for a branch. Someone please give me a time estimate for how long it will take to implement all the features (bugs aside). We can make a branch for a security release after the fact, either off the last release or off the pre-auth-change tag and then build a security-fix-only release off that branch. Why would we branch for a security fix in lieu of just branching a sandbox for the auth developers to play in -- in the first place? -aaron
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 04:51:56PM -0700, Aaron Bannert wrote: ... We can make a branch for a security release after the fact, either off the last release or off the pre-auth-change tag and then build a security-fix-only release off that branch. Why would we branch for a security fix in lieu of just branching a sandbox for the auth developers to play in -- in the first place? Quite simple, actually. You put the developers in definite pain by making them branch, compared to a *potential* pain of a branch for a security fix. pain being arguable, of course -- it all depends upon people's aversion to CVS branches. But there is real pain in the sense that a person will be working by themselves, rather than in the trunk where others will verify the work being completed. And in any case, a branch for a security fix will most likely be done against the 2.0.40 tag, rather than the head. People are going to patch against 2.0.40, if anything. And, honestly, I think people are simply way to frickin' scared here. You should stop and look at the patches that Dirk and Justin have written and are proposing. There is existing code for this. It is mostly *other* people who are talking about destabilizing. Not Justin. Cheers, -g -- Greg Stein, http://www.lyra.org/
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 04:26:22PM -0700, Justin Erenkrantz wrote: aaa rewrite belongs in: 2.0: rbb, brianp, dreid, gstein, jim, rederpj, striker, trawick, ianh, gs, bnicholes 2.1: dpejesh, chris, aaron, hb If someone would like to do a release before I check in the aaa changes, I'd be game to hold off until Monday or Tuesday. (I don't have the time to be an RM right now.) -- justin I would actually like to see a 2.0.41 release. Subversion is seeing problems with working properly against 2.0.40 because of the apr_errno.h changes. If we get a 2.0.41 out there, then we can tell people at least 2.0.41 rather than forcing them to use CVS or a CVS snapshot. I'll try to get the release done over the weekend. Hell... at least a tag. Given a tag, then you can do whatever the heck you want. Cheers, -g -- Greg Stein, http://www.lyra.org/
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 05:22:49PM -0700, Greg Stein wrote: I'll try to get the release done over the weekend. Hell... at least a tag. Given a tag, then you can do whatever the heck you want. This is fair as long as it doesn't unreasonably disrupt anyelse's ability to work on whatever they want to work on as well. -aaron
Re: [TALLY] aaa rewrite
On Thu, Sep 05, 2002 at 05:21:34PM -0700, Greg Stein wrote: Quite simple, actually. You put the developers in definite pain by making them branch, compared to a *potential* pain of a branch for a security fix. pain being arguable, of course -- it all depends upon people's aversion to CVS branches. But there is real pain in the sense that a person will be working by themselves, rather than in the trunk where others will verify the work being completed. And in any case, a branch for a security fix will most likely be done against the 2.0.40 tag, rather than the head. People are going to patch against 2.0.40, if anything. And, honestly, I think people are simply way to frickin' scared here. You should stop and look at the patches that Dirk and Justin have written and are proposing. There is existing code for this. It is mostly *other* people who are talking about destabilizing. Not Justin. The way I see it, Justin is asking for an exception to the rule. The only reason in my mind that this exception was even considered was because it seems logical that we use CVS for what it's good at -- cooperative development. Under normal circumstances I would simply expect the proposed changes to be written up in a patch, posted to the list, reviewed and, if they seem to do their job, committed. Since the full-featured patch hasn't been posted, I have only to assume that this will immediately break the server until they get around to fixing it. I think my stance has been quite reasonable, in that if that destabilizing period is expected to go on for over a week, that the developers interested in this please make a branch. If they do not wish to do this, then they're going to have to follow the normal rules of development here, and do the development on their own and only commit it once they believe it to be correct. -aaron
