On Mon, Oct 3, 2011 at 10:15 PM, Brion Vibber brion@wikimedia.org wrote:
I would *very* strongly recommend doing the internal refactoring before we get anywhere near reviewing and deploying that bad boy; otherwise we'll spend all the code review time pointing out things to refactor to avoid future maintenance problems. :)
I agree. The only question is whether the current system is running out of steam badly enough to require a two stage process. If the answer is "no, we're good", then definitely FileStore first.
Στις 03-10-2011, ημέρα Δευ, και ώρα 22:21 -0400, ο/η Russell Nelson έγραψε:
On Mon, Oct 3, 2011 at 10:15 PM, Brion Vibber brion@wikimedia.org wrote:
I would *very* strongly recommend doing the internal refactoring before we get anywhere near reviewing and deploying that bad boy; otherwise we'll spend all the code review time pointing out things to refactor to avoid future maintenance problems. :)
I agree. The only question is whether the current system is running out of steam badly enough to require a two stage process. If the answer is "no, we're good", then definitely FileStore first.
Weeell... could you give an estimate on the time required for the refactor, given the notes on the page so far, vs. an estimate for time to roll-out without? I admit my original dream was that the refactor would occur *first* and that the Swift backend would be written against that. Ah well, dreams are so ephemeral...
Ariel
http://wikitech.wikimedia.org/view/Media_server/FileStore has a first-cut list of file operations as they map to existing functions. It seems do-able. Mostly a matter of creating another class and defining the semantics of the methods in detail. Probably take me a month and a half (will have other work, hackathon, and Hackers Conference), between design, writing test cases, and coding.
On Tue, Oct 4, 2011 at 1:11 AM, Ariel T. Glenn ariel@wikimedia.org wrote:
Στις 03-10-2011, ημέρα Δευ, και ώρα 22:21 -0400, ο/η Russell Nelson έγραψε:
On Mon, Oct 3, 2011 at 10:15 PM, Brion Vibber brion@wikimedia.org wrote:
I would *very* strongly recommend doing the internal refactoring before we get anywhere near reviewing and deploying that bad boy; otherwise we'll spend all the code review time pointing out things to refactor to avoid future maintenance problems. :)
I agree. The only question is whether the current system is running out of steam badly enough to require a two stage process. If the answer is "no, we're good", then definitely FileStore first.
Weeell... could you give an estimate on the time required for the refactor, given the notes on the page so far, vs. an estimate for time to roll-out without? I admit my original dream was that the refactor would occur *first* and that the Swift backend would be written against that. Ah well, dreams are so ephemeral...
Ariel
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org