my total-expire annoyance: shadowing

my total-expire annoyance: shadowing

[Justin Sheehy]

So, I recently posted about how total-expiry was just doing nothing at
all for me.  I discovered what the problem was:

I had an old copy of make-regexp.el in my load-path, which defines the
function (time-since), among other things.  This was shadowing the
copy of this function in time-date.el and being used by
(nnmail-expired-article-p) in Gnus' expiry process.  The version in
make-regexp.el expects a totally different sort of argument than the
one defined in time-date.el, which would normally have caused an error
to be thrown.  This would have alerted me to the problem much earlier.

However, (nnmail-expired-article-p) calls this function inside a
condition-case, so it simply returned nil instead of throwing an
error.  This resulted in Gnus never thinking that articles were old
enough to expire.  Ugh.

The functionality that I liked from make-regexp.el has long been
superseded by regexp-opt, so I simply got rid of the library.  Once
nnmail started calling the right (time-since), everything was fine.

That was rather annoying.  Why is that condition-case there?

-Justin

Re: my total-expire annoyance: shadowing

[Lars Magne Ingebrigtsen]

Justin Sheehy <justin@linus.mitre.org> writes:

> That was rather annoying.  Why is that condition-case there?

Because the parameter to time-since can be a malformed date string,
which will (in some cases) throw an error of some sort.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen