MZMcBride <z(a)mzmcbride.com> writes:
Bugs have a habit of setting their own priority. If
these bugs were
regularly being hit by users, they would have been resolved ages ago (in
Since discussing this on-list last week, someone pointed me to bug #2700
where Roan speculates a year ago about the possible cause of this sort
What I think happens is <tags> get armored before PST; <nowiki>
needs this treatment, but for other tags it causes this bug.
But they only appear in very, very strange edge cases,
which have the
lowest user impact (not quite zero, but slightly above).
If Roan is right, then this sort of thing has to be worked around all
the time. In fact, if you look at the the most recent comment on #2700
(made tonight), you'll see the troubles of an extension author:
Another comment tonight (this one quite lengthy) on another bug tonight
shows that this sort of problem isn't quite zero.
My recommendation would be to add some parser tests
for these bugs, mark
them as failing, and if the rewrite doesn't take care of them, re-examine
I like the idea of gathering these failures together. Perhaps if they
were grouped then we could see the sort of things that experienced
wikitext writers just don't do, because they are familiar with the
There are so many more bugs that are affecting so many
The fact that you can't combine three types of esoteric wiki syntax in
certain ways currently really shouldn't be a very high concern.
Look at the number of duplicates that #2700 has, the most recent from
this past April. And then there is the number of possible duplicates
that I'll speculate haven't been filed because people were told “Oh, you
can't do that” by someone more experienced than them.
Which reminds me: now I am going to work on that report of duplicates
that you suggested a few months ago now