334747 - VersionCheck.php producing "update.rdf" that points at old version

Status: VERIFIED FIXED

(addons.mozilla.org Graveyard :: Public Pages, defect)

Description

[Mike Shaver (:shaver emeritus)]

The JetEye extension (https://addons.mozilla.org/firefox/1508/) has 3 versions (shown in the order presented on the site, which is not the true 'version' order!):

   0.9.0.20051121115044
   0.10.0.20060418201036
   0.10.0.20060417233625

When a user has the oldest version installed, Firefox requests update info with the URL in the URL field above, but the result points at the currently-installed version, not the most recent update (or another, newer update).

This obviously prevents the extension update from being offered, which is a pretty serious issue!

Comment 1

[Mike Shaver (:shaver emeritus)]

Query used for that URL:

        SELECT
            main.guid AS extguid,
            main.type AS exttype,
            version.version AS extversion,
            version.uri AS exturi,
            version.minappver AS appminver,
            version.maxappver AS appmaxver,
            applications.guid AS appguid
        FROM
            main
        INNER JOIN version ON main.id = version.id
        INNER JOIN applications ON version.appid = applications.appid
        WHERE
            main.guid = '{d07a4843-111f-4699-8551-8ce2afa075cd}' AND
            applications.guid = '{ec8030f7-c20a-464f-9b0e-13a3a9e97384}' AND
            (version.OSID = 1  OR version.OSID = '3'  ) AND
            version.approved = 'YES' AND
            '1.5.0.2+' >= version.minappver AND
            '0.9.0.20051121115044' <= version.version
        ORDER BY
            version.vid DESC
        LIMIT 1

The issue here seems to be the sort order of "0.9.0" vs "0.10.0", and we can just remove the last clause in question, since we'll always want the one with the highest version.vid.

morgamic's going to do that when he gets to a computer (several hours).

Comment 2

[Wil Clouser [:clouserw]]

Patch is in CVS, a.m.o update requested.

Comment 3

[alanjstr]

Was this filter there so that it didn't bring back earlier versions?

Comment 4

[Michael Morgan [:morgamic]]

It was there because it wasn't deleted when it should have been.  (my fault)

Comment 5

[Michael Morgan [:morgamic]]

AMO BUGSPAM FOR COMPONENT MOVE AND DELETE (FILTER ME)

Comment 6

[Reed Loden [:reed]]

AMO BUGSPAM FOR QACONTACT FIX (FILTER ME)