This is closely related to bug 24156 but is an independant issue.
Currently, log messages referring to a user name - usually the
user who caused the log entry to be created by doing something -
list the user name in a quasi tabular format in front of the
message telling what exactly happened.
In English, this can be read as "user-name caused this and that."
giving a nice sentence. Messages are worded accordingly. This
does not translate to languages, where "user-name" cannot be
the first word in a sentence. (Grammar, syntax, word order,
good taste or style, and politeness rules may be prohibiting)
It might thus be advisable, to tie the user name into the
message text so as to allow it to appear at any place in a
sentence.
This gives up, however, some of the ease of machine processing
coming with its current fixed position. My current opinion is
that machine readable logs should be availabe much better via
the API, and fixed positioning should be given up in favour of
better i18n and language support.
The possible half-way solution, keeping the tabular user name
and passing the user name to the message handling of the textual
part, would allow genderized messsages, and the message could
even repeat the user name instead of using "this user" etc.,
but I think that makes log entries less readable.
We could of course include the user names in the message texts
andkeep the tabular user name field, assigning it to a
display:none CSS class. This increases log page size.
Yet this dupliction, without display:none, may be a good interim
solution while messages are beeing adopted to the new style.
Another way to make a transit could be to identify all
messages being used for logging, insert "[[user:$1|$1 "
at their beginning in all message source files, and alter
message handling function calls accordingly. Thus, translators
had unlimited time to fix them.
Version: 1.20.x
Severity: enhancement