Thread:User talk:Siebrand/Question (6)/reply (13)

Agreed with Nike, we could document at best half of the messages in MediaWiki and its extension in an automated way, telling whether or not they accept html inline markup, and whether or not they are being used in an html context. I tried to develop a simple text based xref tool which would be able to list which message key is/was used where (per file name and revision number) in the subversion repository, which could also collect some usage type information. I gave up because even identiying message keys is often impossible without "understanding" the "meaning" of program code. For example $abc='nuke' in one place and $messagestring=wfMsg($nuke.'-desc') in another is too hard for a text based analyzer program looking for occurrances of a message key 'nuke-desc', and we have too many more complicated cases than this.

It is worth noting usage informations in the message documentation, when it is obvious that it will not change, such as "this the body of an e-mail sent to users when…" - the use as an e-mail body implies that html must not be used.

Else, developers alter things like whether inline html makup is accepted or not. I do, for example. There are places where html inline markup is unproblematic but not (yet) supported, and others where a change would be either insecure or call for large amounts of work. If you find instances of messages where you would recommend using html markup and it does not work, let us know via the support page. Odds are, that a change would be simple and doable. Inside tabular structures and inside links and headlines that is sometimes not so, or at least not easy. Filing a bug at buzilla may alert developers who know the particular piece of code well enough to make the change. If you file such bugs, please mark them with the keyword i18n so that we more easily find them.