[qmtest] Why is qmtest installed in such a non-standard way?
Nathaniel Smith
njs at pobox.com
Sat Jan 4 06:45:33 UTC 2003
[Apologies again if this is a double post; I think ezmlm hates me.
But I've tweaked my mail config a bit, and perhaps that will pacify
it...]
On Thu, Jan 02, 2003 at 09:23:00AM -0800, Mark Mitchell wrote:
> --On Thursday, January 02, 2003 12:34:01 AM -0800 Nathaniel Smith
> <njs at uclink4.berkeley.edu> wrote:
>
> >QMTest installs rather differently from every other Python library
> >I've used; in particular, it uses a non-standard location, and
> >requires some quirky environment variables to be set before it can be
> >used.
>
> That's probably a fair criticism, but QMTest is an application, not a
> library. I think it's reasonable to ask it to do a better job at being
> a library. We haven't done this because, well, it takes effort. :-)
Hopefully it doesn't take _too_ much effort though; one of the nice
things about python IMHO is that every application is also a library,
almost automatically. The only thing preventing this happening for
QMTest is the non-standard install.
Granted, poking around a bit, the environment variables seem to have
snuck into a number of different places; I'm not sure exactly how one
would best get rid of them. I'd be happy to spend some time working
on a patch, though, if you could provide some idea of the required
behaviour -- it seems like a lot of the quirkiness has reason to it
that I just don't understand at the moment :-).
> Part of the reason for using XML as the default database format is to
> allow people to build databases from any XML-generating tool -- including
> ones that are *not* written in Python.
Yes, I'm aware of this (in fact, I think I suggested to Zack a few
months ago that freezing and documenting the XML structure would be a
good idea); on the other hand, I'd like to be able to support any of
the alternative database implementations that you recommend people
develop, and besides. Besides which, the advantage of XML is really
that it makes it easier to reimplement the default database class in
your language of choice; I think, though, it would be even easier than
that to just use the implementation that _already_ exists in my
language of choice :-).
And, of course, there are things besides the database code that would
be useful; qm.label comes to mind as an obvious example, as does
playing with my new test classes interactively (which desire is what
actually got me started unraveling qmtest's environment variables).
I'm probably beating a dead horse by this point, so I'll stop.
Regarding changes to the install, what needs to be done?
-- Nathaniel
--
The Universe may / Be as large as they say
But it wouldn't be missed / If it didn't exist.
-- Piet Hein
This email may be read aloud.
More information about the qmtest
mailing list