[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Meta-test suite

On Wed, 9 Mar 2005, Per Bothner wrote:

> Donovan Kolbly wrote:
> > I've started working on a meta-test suite for SRFI-64.
> Cool!  It would be nice to incorporate such a test into SRFI-64 itself,
> if you agree.

That would be great by me.  I've always thought that test suites should be
a standard part of SRFIs, and now, with SRFI-64, we have the foundation to
start doing so.  What better place to start?

> I'm really overloaded with various things, but I'll try it out asap.

That'd be cool, if you get a chance.  The latest one tickles some
ambiguities in the spec, too.

> I'd alse like to see the test-runner you used to generate the web page.
> Is that portable?  I assume it should be possible to make a portable
> version, by just display-ing html directly.  (I can see the line
> number links might not be portable.)  If so, it might be nice
> to also make that available from the SRFI document, at some point.

It's not portable at all, although that would be nice.

Currently, it's generated in two steps.  I have a test-runner which
generates an XML summary (using SXML internally and then writing out XML)  
[source code:  
and then the web server converts it into HTML on the fly (again, mostly
SXML crunching, but there also using RScheme's xpath facility).

As you say, the line-number stuff is highly non-portable (I even had to 
drop a quick hack in RScheme to get the line # that a form *ends* on), but 
the runner in theory could just extract that using the test results alist.  

More crucially, there does not seem to be any way for a portable runner to
keep track of the nesting of test suites.  I'm not sure yet what to
suggest for SRFI-64 to track suite nesting, but something along the lines
of a `test-runner-on-suite-begin!' and `test-runner-on-suite-end!' would
probably do it (*).

In my implementation, the default runner keeps track of all results all
the time, complete with nesting structure, and so when I'm dumping the
results it's easy to walk the test-result tree using my internal data

[*] do you want to call these things "suites" or "groups"?  The
    terminology seems to vary from place to place, or am I still missing

-- Donovan Kolbly                    (  d.kolbly@xxxxxxxxxxx
				     (  http://www.rscheme.org/~donovan/