Ronald Spengler wrote:
> Ron the bug collector says: > > We should still collect bugs, and have a place to put them, while we're > studying them, but we shouldn't keep them very long, as they've lots to > do, and little time. I agree. But I will say that I much prefer contributions that go into the inbox and for two very simple reasons: 1) They have context. I don't need to worry whether the change conflicts with other changes that have happened in the meantime - Monticello takes perfectly care of that. Contrary to Mantis where you can find fixes filed out from every Squeak version under the sun (just look at the preamble of most fileouts) you know the definitive (package) version this patch was made against. You also know that whoever posted the fix must have had an image that wasn't broken beyond repair; at the very least they were able to post it. And you know that at least it has been loaded into a Squeak that was reasonably similar to yours (based on package version). All of this knowledge about the patch is immensely useful for integration. 2) It is an *extremely* efficient workflow. I can fire up Squeak, click on the inbox repository, open it. Within three clicks I can see if there has any actual code been contributed. It takes a lot more work than that to find actual code on Mantis. Then there is loading. I have Monticello already open, clicking on the package on "changes" or "merge" allows me to review the actual code against the current version within two more clicks. Compare this to Mantis where you'd download the fix to the directory, open a file list, click on the file, choose "browse changes". After loading and verifying the change there is posting it to the trunk. Monticello is still open, unless there was a issue or I needed to merge I can simply copy the version into the trunk with two more clicks. With Mantis, I load the change, then I need to find which packages are affected by it, then I need to write change log comments (possibly copying them out of the change set preambles etc). In other words, the integration workflow with Mantis takes me somewhere between 10-15 minutes even for trivial fixes, including usages of multiple tools that aren't integrated. With Monticello and the inbox the same basic process takes me seven clicks. I've never felt anywhere near as effective for integrating stuff. Cheers, - Andreas |
In reply to this post by Randal L. Schwartz
Am 02.09.2009 um 14:30 schrieb Randal L. Schwartz: >>>>>> "Bert" == Bert Freudenberg <[hidden email]> writes: > Bert> Invisibility of the actual development process was a big > problem. I'd > Bert> rather have too much visibility than the impression that there > is no > Bert> work being done. > > I've had people on IRC comment "I didn't realize there was so much > activity on > the core" regarding the emails. And the squeaksourcebot showing off > the > 50-150 commits a day to squeaksource in general is also good to show > that > there are as many packages being uploaded to Squeaksource every day > as there > are Perl CPAN modules, which we've often held as a sign that Perl > activity is > alive and well. I think having many commit messages is a great sign for our community. They are a great way to track what is being done in Squeak. And please don't move them to another mailing list! We have already too many of them which increases fragmentation which is IMO one of the biggest challenges for the Squeak - and more general Smalltalk - community. I would kindly ask those who don't like the commit messages to use their mail client's features to handle them as they wish. Cheers, Bernhard |
I just want to echo the approval for the commit messages. I was just able to harvest Nicholas' changes for ScaledDecimal literal correctness. Very very nice (no pun intended).
Strikes me that one way we could reduce the size of the commit messages is by modifying the diff algorithm to output missing or added methods not as the source with '+''s or '-''s at the start of the line, which is verbose and tedious as one has to scan all lines to verify they all begin with '+' or '-', but as "Added ClassFoo>>selector" & "Removed ClassFoo>>selector". Any volunteers? If not, where does the diff code live and I'll try and get it done.
best Eliot
|
On 12.09.2009, at 22:54, Eliot Miranda wrote: > I just want to echo the approval for the commit messages. I was > just able to harvest Nicholas' changes for ScaledDecimal literal > correctness. Very very nice (no pun intended). > > Strikes me that one way we could reduce the size of the commit > messages is by modifying the diff algorithm to output missing or > added methods not as the source with '+''s or '-''s at the start of > the line, which is verbose and tedious as one has to scan all lines > to verify they all begin with '+' or '-', but as "Added > ClassFoo>>selector" & "Removed ClassFoo>>selector". Any > volunteers? If not, where does the diff code live and I'll try and > get it done. http://source.squeak.org/ss.html Package SqueakSource Class MCDiffyTextWriter - Bert - |
Free forum by Nabble | Edit this page |