RE: subversion client test suite

I believe Bob was recalling a test harness from our
collective past that did what he said as an option.

That is, ordinarily, the test harness ran everything
within its own framework, connecting to a database
that had the test information, etc. However, the
harness had the option to "export" the test suite
to a collection of files and shell scripts so that
you could run the suite on a minimally-configured
standalone machine. This was especially useful for
performance test suites that generally shut everything
possible down before executing in order to minimize
external effects on the performance of the system.

The harness was known as "TD" and later as "CITE"
(Convex Integrated Test Environment) and was documented
in a few papers I wrote back in the early 90's for
ISSTA and PNSQC. Alas, I have no idea what happened
to this harness after HP baile..., er, sorry, "assimilated",
no that's not right either, um "acquired" that's it
Convex in late 1994.

It was actually a pretty cool harness for its time, though
if I were to build a harness now I would probably embed
Perl into it, rather than the test-specific language
we'd designed then. Perl and Python are great languages
for this sort of thing, personally I prefer Perl, but
that's because I've used it for 12 years (and yes, I
*can* pick up a program I wrote 12 years ago and know
what I was doing, but then, I try to write readable
code, not use the latest whiz-bang ultra-dense
one-liner perl that many perl people tend to extoll).

-Peter

--
Peter A. Vogel
Manager, Configuration Management
Arsin Corporation
4800 Great America Parkway Suite 425, Santa Clara, CA 95054
> -----Original Message-----
> From: Lee Burgess [mailto:lefty@red-bean.com]
> Sent: Thursday, March 08, 2001 9:01 AM
> To: Bob Miller
> Cc: dev@subversion.tigris.org
> Subject: Re: subversion client test suite
> 
> 
> Bob Miller writes:
>  > Mo DeJong wrote:
>  > 
>  > > Same goes for automake, yes it is written in Perl but it does not
>  > > require Perl at build time, it is only required on the developers
>  > > machine.
>  > 
>  > Could the subversion test suite use this approach?
>  > 
>  > Require a real scripting language to create Bourne shell 
> test scripts,
>  > then ship the test scripts in the source tarball.
>  > 
>  > Developers will need the real scripting language, and 
> installers will
>  > need Bourne shell.
>  > 
>  > Best of both worlds, and only adds one more level of hairy 
> complexity
>  > to the test harness design.  (-:
> 
> I am hesitant to implement such a system.  However, I could possibly
> be convinced otherwise.
> 
> Mainly, I don't like the idea of writing a framework in Perl (or
> Python or Tcl) which is really generating test suite code in Bash.  I
> think it is again overkill for the purposes of the subversion client
> test suite.
> 
> Please note also, that the client test suite is something that
> operates independently of the test harness design.  I could be wrong,
> but that is what I gathered from my discussion with Karl and Ben.
> 
> -- 
> Lee P. W. Burgess  <<!>>  Manipulate eternity. Power is a symphony:
> Programmer         <<!>>  elaborate, enormous, essential.
> Red Bean Software  <<!>>  Dream the moment with a fiddle in summer 
> lefty@red-bean.com <<!>>  and a knife in winter.
> 
>