Page MenuHomePhabricator

Remove references to code deprecated before 1.17 in 1.18
Closed, DeclinedPublic

Description

$wgAllowUserSkin/$wgAllowRealName/$wgExtensionAliasesFiles/$wgExtraRandompageSQL

Linker::tooltip() (from 1.14!)

since 1.16. Most are calling each other, so can be generally cleaned up
Linker::makeBrokenLink()
Linker::makeLinkObj()
Linker::makeKnownLinkObj()
Linker::makeBrokenLinkObj()
Linker::makeColouredLinkObj()
Linker::makeImageLinkObj()

LogReader and LogViewer but that's bug 24711

Seems the Linker ones are the bigger issues...


Version: 1.20.x
Severity: minor

Details

Reference
bz28863

Event Timeline

bzimport raised the priority of this task from to Low.Nov 21 2014, 11:37 PM
bzimport set Reference to bz28863.
bzimport added a subscriber: Unknown Object (MLST).

$wgExtensionAliasesFiles is another

makeBrokenLink is called by Interlanguage Extension

makeLinkObj is called by AjaxShowEditors, FCKeditor and SemanticForms

makeKnownLinkObj is called by SemanticForms (x2) and in Linker::makeColouredLinkObj

makeBrokenLinkObj is called by Linker::makeBrokenLink

makeColouredLinkObj is called by Linker::makeColouredLinkObj (deprecated in 1.17)

makeImageLinkObj killed

(In reply to comment #1)

$wgExtensionAliasesFiles is another

I think we should keep that around a little bit longer so extensions don't break back-compat. Maybe next release.

It should be fine to start removing the code for 1.18

moving to low priority since we're not desperate to get this done.

Not a 1.18 blocker, and one that we shouldn't do for the sake of doing. If there's a feature that we're working on that breaks these, then we can remove them as part of that work, but we shouldn't break backwards compatibility just for giggles.

(In reply to comment #6)

Not a 1.18 blocker, and one that we shouldn't do for the sake of doing. If
there's a feature that we're working on that breaks these, then we can remove
them as part of that work, but we shouldn't break backwards compatibility just
for giggles.

Well stuff that's already marked for deprecation can be slowly phased out once extensions have all been updated (and then the general consensus is to wait at least one more release). When we mark stuff @deprecated these days we generally mention when they were first deprecated so we can properly remove them over time.

But yes, removing deprecated code shouldn't be a blocker unless it's breaking something by remaining there.

No real reason for this to explicitly stay open