>
> We should investigate how these 2 projects can work together when I'm
> done so that RoboWeb 'recorded' sessions can create WWW::Automate methods
> that utilise the structure of the page. The benefits of doing this are:
> 1- Resultant scripts are easier to interpret
> 2- Scripting apps that vary their form field names works.
>
RoboWeb will reproduce exactly what was sent by the browser during the
recorded sequence, except the for the cookies, which are handled
independently each time the sequences are run. This can be limiting because
depending on database state many applications will generate different URLs
and hidden form parameter values each time a page is visited, which would
render RoboWeb tests unusable.
I have overcome this by adding infrastructure to put the database in
'test-scenario state' just before each of the sequences in the suite is
recorded and before each of them is run. This approach works fine for most
cases but may be impractical if the database to be restored is large and is
also not suitable to testing error conditions that may develop in the
database after some use. That databases are restored also makes it
unsuitable for use on production sites.
So yes, number 2 in your list is something that would be great to have. To
that I would add A) the capability of extracting hidden parameters
automatically and use those instead of the ones originally recorded. B) To
generate requests based on urls in the page instead of the ones in the
recorded sequence. This would render the database restoring step
unnecessary, but would certainly require more human intervention when
recording the test sequences.
I look forward to seeing WWW::Automate and to working together on this.
Claudio
>
> --
> Jeremy Howard
> [EMAIL PROTECTED]
