Page MenuHomePhabricator

maintenance script to populate rev_len field for old revisions
Closed, ResolvedPublic


Author: matthew.britton

This would be useful for various stuff, mostly along the lines of statistics generation, where currently the whole text of every revision has to be fetched, which is obviously quite inefficient and resource-heavy.

It might be too much work; Brion said "it's feasible, but you shouldn't assume it'll be done on any particular installation", but I can't figure out if that means "no" or not. :)

Version: unspecified
Severity: enhancement



Event Timeline

bzimport raised the priority of this task from to Medium.Nov 21 2014, 10:43 PM
bzimport set Reference to bz18881.
bzimport added a subscriber: Unknown Object (MLST).

Updating product/component. wrote:

There's no reason to restrict it to enwiki, even if it does have more revisions than all other WMF wikis put together...

Changing summary so all who wish can have and run such a script.

Ah, finally the older part of action=history lists needn't look brain damaged: "why no sizes?"

Maybe make it called by update.php too?

Say, some maintenance scripts are only needed once. Perhaps this
should be put in a sub directory.

matthew.britton wrote:

Changing summary as the scope of my original request seems to have been expanded from "english Wikipedia" to "all mediawikis", so the cutoff date will depend on when they upgraded.

Changing Product: Mediawiki → Wikimedia
Changing Component: Maintenance Scripts → Site Requests
Changing Keywords: (N/A) → Shell wrote:

No, the problem is that there *is* no such script. Someone needs to *write* one before it can be run on WMF.

(In reply to comment #6)

No, the problem is that there *is* no such script. Someone needs to *write*
one before it can be run on WMF.

This bug is written in the pose of running the script, 12188 is the one requesting the script (in which there is already one patch). wrote:

If anything (and per bug12188 c6 by Brion) this is the 'write the script' bug, and bug12188 is the 'run the script' bug. Or you could say this is a dupe, I guess. Regardless, the best way to proceed is to actually write the damn script :D

Added maintenance/populateRevisionLength.php in r63221.