There's probably a lesson to be learned here from the Extreme Programming guys, about aligning authority and responsibility.
It's frustrating for me to feel like I have the responsibility to review a changeset when I don't have the authority to put that changeset into the image myself.
And I *do* feel like I have that responsibility, because I felt guilty just now when Marcus asked us how we can possibly expect the harvesting process to work if we submit changesets but don't help with the reviewing.
I think that the answer to his question, though, is that the way it *should* work is that when someone submits a changeset affecting the Wobulator package, the owners (stewards?) of the Wobulator package should be the ones responsible for reviewing it - because those are the people who have the authority to get it into the official Wobulator distribution. I haven't reviewed any non-Celeste-related changesets in a long time, but I try to catch all the Celeste-related ones.
This is why I've been doing the package-removal work, but there's no reason why we can't do this for code that's still in the image (as long as the guides/harvesters/whoever are willing to let a package's steward mark changesets related to that package as [APPROVED]). Isn't that why we were looking for stewards?
http://minnow.cc.gatech.edu/squeak/3088
Adam
______________________________________________________________________ Post your free ad now! http://personals.yahoo.ca
On Nov 2, 2003, at 2:08 PM, Adam Spitz wrote:
I think that the answer to his question, though, is that the way it *should* work is that when someone submits a changeset affecting the Wobulator package, the owners (stewards?) of the Wobulator package should be the ones responsible for reviewing it - because those are the people who have the authority to get it into the official Wobulator distribution. I haven't reviewed any non-Celeste-related changesets in a long time, but I try to catch all the Celeste-related ones.
I absolutely agree with this. Marcus asks what has happened between 3.2 and 3.6. The answer is that we have come a long way (not, by all means, all the way) towards breaking the image into separately maintainable packages, which will eventually allow an effective distribution of the reviewing and harvesting tasks. IMO, that's the only way harvesting is going to work - with changesets submitted by everyone, but reviewed and incorporated by package maintainers.
Avi Bryant wrote:
On Nov 2, 2003, at 2:08 PM, Adam Spitz wrote:
I think that the answer to his question, though, is that the way it *should* work is that when someone submits a changeset affecting the Wobulator package, the owners (stewards?) of the Wobulator package should be the ones responsible for reviewing it - because those are the people who have the authority to get it into the official Wobulator distribution. I haven't reviewed any non-Celeste-related changesets in a long time, but I try to catch all the Celeste-related ones.
I absolutely agree with this. Marcus asks what has happened between 3.2 and 3.6. The answer is that we have come a long way (not, by all means, all the way) towards breaking the image into separately maintainable packages, which will eventually allow an effective distribution of the reviewing and harvesting tasks. IMO, that's the only way harvesting is going to work - with changesets submitted by everyone, but reviewed and incorporated by package maintainers.
I second this...a package maintainer is the best person (or persons) to review and accept changes. Unfortunately, I don't think we have package maintainers for most of the components in the base Squeak image. Perhaps it would help if we had a couple or three maintainers assigned for each subsystem in the base Squeak. It might also help if bug/fix submissions were categorized by package. It would also be nice if we used something other than the mailing list for reporting and tracking bugs and enhancements to packages.
- Stephen
It's frustrating for me to feel like I have the responsibility to review a changeset when I don't have the authority to put that changeset into the image myself.
Hi adam
I think that there are several points to consider: - first if you review a changeset and say that this is well implemented fix a real problem, works well, then the harvester (the guy that has the responsibility not to break the system) will just have a quick look and this will be done
- second there is a question of trust. I would not like to have a system in which any people can add something that can break the rest. Believe working on KCP is sometimes getting on the nerves.
So I think that marcus wanted to say that we need - more people to give their point of view - more harvester in specific domain: network, Morphic, concurrency,...
And I *do* feel like I have that responsibility, because I felt guilty just now when Marcus asked us how we can possibly expect the harvesting process to work if we submit changesets but don't help with the reviewing.
I think that the answer to his question, though, is that the way it *should* work is that when someone submits a changeset affecting the Wobulator package, the owners (stewards?) of the Wobulator package should be the ones responsible for reviewing it - because those are the people who have the authority to get it into the official Wobulator distribution. I haven't reviewed any non-Celeste-related changesets in a long time, but I try to catch all the Celeste-related ones.
This is why I've been doing the package-removal work, but there's no reason why we can't do this for code that's still in the image (as long as the guides/harvesters/whoever are willing to let a package's steward mark changesets related to that package as [APPROVED]). Isn't that why we were looking for stewards?
But I would dream about that but we need a good package dependencies mechanism before doing that frankly. We should be able to say Squeak 3.7 is built from xxx7 vv3 bb3.1 and so on.
Stef
http://minnow.cc.gatech.edu/squeak/3088
Adam
Post your free ad now! http://personals.yahoo.ca
squeak-dev@lists.squeakfoundation.org