I would like an enhancement to megatest:
A) A setting under megatest.config that allows the definition
of a pre-execution script/hook/whatever that will run early
on under a minimal megatest environment (i.e. has MT_* vars
defined). For example:
[checks]
before_start = <script & options to call before regression start>
before_launch = <script & options to call before test launch>
after_launch = <script & options to call after test launch>
You get the idea. The script may be the same in all cases, just with
options to identify the "phase/context" its being called under, for
example:
[checks]
before_start = checkit.rb --start --pre
before_launch = checkit.rb --launch --pre
after_launch = checkit.rb --launch --post
It doesn't really matter, just so long as the script is designed
and allowed to return logpro-style exit codes (0, 2, 3, 4, 5), and
print text to stdout that will be captured by megatest for display.
The [checks] section should allow for easy extension of other additional
check hooks over time if necessary.
B) A method in megatest GUI to pop-up a dialog box with the text printed
by the script, the error code reported (unless its PASS or WARN), that
requires the user to click either "I'll fix it", or "Proceed".
C) The results of any check script should appear under logs/ using the
same name as the check, for example: logs/before_start.log. Since the
checks are done on a per-execution basis, any new execution will overwrite
existing logs, so users need to deal with the issue recorded before
running again (and overwriting the logs).
D) in non-GUI mode when running megatest from the command line, any result
other than PASS/WARN should cause megatest to stop execution unless the
user decides to rerun with the -proceed option (which is a fallback
override in case the check scripts have problems). But even so, if they
decide on this options it will be very clear to user and to us that they
have done so despite the reported problems.
|