> 1) XML isn't nice to read -- and doesn't really belong on an
> output window
> If the end goal is to reduce clutter out to the user when running (like not
> seeing each and every file get copied), then I'd rather see more effort
> being placed into logleveling the messages with the default loglevel seen
> set pretty high.
The end goal is not to reduce clutter, but to allow post-processing.
If every task report it's messages in it's own way.
Think about tools - ant is extensible and that means it will be very hard
to automate it ( not only "nightly", but
also for GUI tools ). Parsing ant output is almost imposible - it's a mix
of various formats, each task has it's own message style and no pattern.
( I know that having ant in a GUI is still far away :-)
The real problem is that we are losing important informations when we log-
and we can't recover the lost. It may be possible to strip the
structure and display plain text, but the reverse is much more dificult.
I know, all tools have text output and (maybe) a parser that understand
the a part of the output ( gcc,javac, sendmail, httpd, etc). Most of the
time it's very hard to work with those tools, and it's much harder to
write code that parse the output produced by arbitrary tasks ( some
probably not even known in advance) than to strip some <> or even read
XML.
Costin
|