On Fri, 10 Jul 1998, Lutz Pressler wrote:
> when domain parts of addresses are compared with the arguments of
> "domains =" or the keys of retry rules they seem to be converted to
> lower case - and then a case _sensitive_ comparison is done.
> That means: the parameters have to be given in all lower case to get the
> expected result!
>
> Is there any reason not to allow mixed case?
You cannot do case-insensitive lookups on DBM files. (Nor, I suspect, on
NIS or other kinds of keyed database.) The function that checks that a
string is in a list does seem to be case-sensitive in its exact matches.
I cannot remember the reasoning behind this, but it might have been so
that all kinds of matching in the list work the same way.
Unfortunately, there is a single function that is used for many
purposes. When handling local parts, it does have to be case sensitive,
because local parts in RFC 822 addresses are specified as being case
sensitive. This leads to a lot of grief, but I don't suppose it will
ever get changed.
Exim does in fact do case-insensitive lookups on lsearch files. I wrote
it like that because it was trivial to do, and meant that people who put
upper case letters in lsearch alias files didn't get caught. The
exim_dbmbuild program that builds DBM files from lsearch files
automaticaly lower cases the keys as it does so.
With hindsight, it might have been better not to hide the casing
problems in this way.
One of the items on my long-term "tidy-up" list is to review all the
case-handling in Exim and try to draw up some general rules and stick to
them. Unfortunately, I don't think this will happen for quite some time.
--
Philip Hazel University Computing Service,
P.Hazel@??? New Museums Site, Cambridge CB2 3QG,
ph10@??? (sic) England. Phone: +44 1223 334714
--
*** Exim information can be found at
http://www.exim.org/ ***
