[MUD-Dev] software engineer: testing (was: [DGN] Creating a MUD)
Miroslav Silovic
miro at vams.com
Thu Jun 6 12:34:24 CEST 2002
Bruce Mitchener wrote:
> If you take the line of reasoning that a lot of stuff that should
> be common sense is allowed to fall into software engineering, then
> we can include things like software testing. :)
> And ... software testing should be done on all projects, large and
> small. Previously, most of the tools that I've liked using were
> fairly expensive (Purify, Quantify), but lately, some pretty nice
> free or cheaper equivalents have come along.
My current development procedures:
- I put a test suite in each source file (I've been using mostly
C++, lately), and use preprocessor to supress them in release
build. The suites generally test the functionality implemented in
each given source file (and when the functionality moves, so do
the tests).
- Tests are automatically run on each application startup, and
assertions make the application bail as soon as any weirdness gets
detected.
- As of late, I add occasional valgrind runs. This valgrinds the
unit tests.
- Each exported interface is documented using doxygen. I write the
docs as soon as the interface gets finalized and implemented
(yeah, I know, I ought to document the interface /before/ I
implement it, but... :) )
- I have strict memory management policies and adher to them
almost religiously.
I do this in my solo projects, too. The effects on the speed of
development have been nothing short of stunning. In particular,
pervasive testing makes me less conservative about redesigning
things - I know my back is covered if I break something (extreme
programming, anyone?).
And also, I can drop some project for a few months, if I don't have
the time,. and resume coding within 10-15 minutes once I decide to
return to it.
Miro
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev
More information about the mud-dev-archive
mailing list