1. Regression Testing
(This message will probably show up once Topica's servers are back up, but I'm
posting it directly to the forum for now.)
I follow the D programming lists even though I don't contribute or really write
anything in D. But the question occurs to me about regression testing.
Before reading the D list I never really knew what regression testing was. I'm
sure that Walter has his own set of tests, but every time something doesn't meet
the published spec, Thomas Kuehne adds it to his project called DStress and
tracks it.
Is there anything formal like this done for Euphoria or Win32lib? How about
other projects? What kind of regression testing is done for Euphoria or Euphoria
projects? How are results tracked?
It's an important question, especially for library projects such as the proposed
standard library.
I'm not really familiar with software testing and that is why I am curious.
=====================================
Too many freaks, not enough circuses.
j.
2. Re: Regression Testing
In respect to Win32lib, one of the reason for so many example/demo programs is
so I can run them all as a sort of regression test. However, it is nowhere near
as organised or comprehensive as I'd like it to be.
I would like at least one demo program for each routine in win32lib, and each
permutation of a Windows control. I'm getting there though.
I think Thomas Kuehne has done a remarkable job with the DStress project. One of
the side-effects that it has is that it gives us a good sense of how stable the D
product is (it is not v1.0 yet)
--
Derek Parnell
Melbourne, Australia
irc://irc.sorcery.net:9000/euphoria
3. Re: Regression Testing
Jason Gade wrote:
> I follow the D programming lists even though I don't contribute or really
> write anything
> in D. But the question occurs to me about regression testing.
>
> Before reading the D list I never really knew what regression testing was.
> I'm sure
> that Walter has his own set of tests, but every time something doesn't meet
> the published
> spec, Thomas Kuehne adds it to his project called DStress and tracks it.
>
> Is there anything formal like this done for Euphoria or Win32lib? How about
> other
> projects? What kind of regression testing is done for Euphoria or Euphoria
> projects?
> How are results tracked?
A good example of a regression test is euphoria\demo\sanity.ex
It quickly and automatically self-tests dozens of features.
I run it and many other programs when I'm adding new code.
I also have a sanity2.ex that I haven't released. I often add new
tests to these programs when a bug is detected.
I also interpret/compile/bind many large programs such as the IDE.
For testing the translator, I save all the .c and .h files in a
directory, and then I do an A vs B diff to see where the
C code has changed. That's a good way of confirming that a
new optimization does what it's supposed to do and
doesn't introduce a bug.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
4. Re: Regression Testing
Jason Gade wrote:
>
>
> posted by: Jason Gade <jaygade at yahoo.com>
>
> (This message will probably show up once Topica's servers are back up, but I'm
> posting it directly to the forum for now.)
>
> I follow the D programming lists even though I don't contribute or really
> write anything in D. But the question occurs to me about regression testing.
>
> Before reading the D list I never really knew what regression testing was.
> I'm sure that Walter has his own set of tests, but every time something doesn't
> meet the published spec, Thomas Kuehne adds it to his project called DStress and
> tracks it.
>
> Is there anything formal like this done for Euphoria or Win32lib? How about
> other projects? What kind of regression testing is done for Euphoria or Euphoria
> projects? How are results tracked?
>
> It's an important question, especially for library projects such as the
> proposed standard library.
>
> I'm not really familiar with software testing and that is why I am curious.
>
> =====================================
> Too many freaks, not enough circuses.
>
> j.
Rob and Derek, thanks for the replies. I didn't think of sanity.ex as a
regression test but looking at the (somewhat complicated) code makes it
pretty clear that that is what it is.
--
==============================
Too many freaks, not enough circuses.
j.
