apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Justin Erenkrantz <jerenkra...@ebuilt.com>
Subject Re: cvs commit: apr/test testlock.c
Date Wed, 06 Jun 2001 23:39:13 GMT
On Thu, Jun 07, 2001 at 12:30:17AM +0100, David Reid wrote:
> Hmm, well all the other test apps are using it...

Doesn't mean they are right.  =)

> The general rule I'm moving to is that the "output" all goes to stdout,
> error messages to stderr.  So if we have
> 
> testing foo
> 
> The OK or Failed are both printed on stdout as part of the "output" and the
> error message as to why it failed goes to stderr.  This means we end up with
> something like
> 
> testing foo            Failed
> we coudn't do foo : [1001] bar is not supported on this platform

As long as we have a convention, I don't care much where things go.
This seems reasonable enough to me.

> The aim is to move to a script based test run whereby we get simple reports
> of what tests passed and what failed with the output being dumped into a
> file, hence the desire for the distinction.  Of course it'll be a while
> before we get there :)  test_apr.h is the start but there is more support
> needed yet.
> 
> If we have to make the change fair enough, but we haven't been using it
> anywhere else in the test apps

Oh, I'd love to see that.  "make check" does the obvious things.  I 
think most tests were written before we had the ability to get stdin, 
stdout, and stderr via APR.

> BTW, do other people even run them anymore?

I do.  -- justin


Mime
View raw message