Re: [Logitest-developer] Features added

[Clancy M. <cl...@cy...>]

On Friday 15 February 2002 03:09, Anthony Eden wrote:
> > -----Original Message-----
> > From: log...@li...
> > [mailto:log...@li...]On Behalf Of
> > Clancy Malcolm
> > Sent: Wednesday, February 13, 2002 9:12 PM
> > To: Anthony Eden; LogiTest Developer
> > Subject: Re: [Logitest-developer] Features added
> >
[snip]
> > Long term I think we need to find a better (preferrably with source
> > code) HTML renderer than both JEditorPane and CalpaHTMLPane which
> > both seem
> > to be limitted.
>
> Agree 100%.  This is the main reason I stopped working on LogiTest - there
> are no open source Java web browser components worth a damn.
>
> > In the shorter term maybe we need to let the user switch
> > between the two so that they can choose to have either recording
> > functions or
> > HTTPS browsing functions.  What are your thoughts?
>
> Seems like a pain.  I would almost rather ignore SSL...My company site uses
> it though (and that was the main reason I wrote LogiTest.)  I am OK using
> CalpaHTMLPane for the moment and just using SSL through explicit resource
> definitions.

OK, I'm happy with that.

> > > - You added DiffTestlet but not to the config.  Is it because
> >
> > you want to
> >
> > > test it some more before adding it to the available testlet list in the
> > > config?
> >
> > It can be added to the config, but it is fairly rough at the
> > moment and only
> > has limitted usefulness.  Part of what is lacking is a way to display the
> > differences between the reference and captured documents.  The
> > only diff code
> > I can find out there is GPL which has the 'viral' effect when you
> > add it to
> > your own code.
>
> Right...that sucks.  I definitely would like a good diff testlet though - I
> am not sure if you have ever used eTest, but that is there method for
> testing success or failure.  It is good that LogiTest lets you define a
> testlet, but it would be nice to be use diff as a default testlet for
> pages.

I'll see if I can find something to do the diffing.  However, for the sites 
that I am testing there is lots of date dependent output (e.g. if you have 
logitest create a task that is due on 2002-02-16 today it will be a regular 
task, but tomorrow it will be a due task).  I found that this made the diff 
testlet unreliable by design.

I haven't used eTest, did they do a character by character diff or something 
with more tolerance?

> [snip]
> > > - The temp file addition to the SaveThread always fails on my
> >
> > machine (and
> >
> > > thus I cannot save).
> >
> > Can you give me a stacktrace?  It seems to work fine for me.
>
> Sure:
>
> java.io.IOException: Could not backup file to
> C:\DOCUME~1\ANTHON~1\LOCALS~1\Temp
> \logitest55082.xml
>         at org.logitest.ui.SaveThread.run(SaveThread.java:91)
>
> Not very descriptive though.  :-)
>

You're right - not very descriptive.  I'll see what I can do.

>
> A couple suggestions as well:
>
>  - The hot key for the start and stop recording is the same...perhaps the
> stop recording button and menu item should be disabled until the record
> button is hit.
>  - It would be nice to have something indicate that you are recording.
> Originally I had a little "RECORDING" flashing in the bottom of the window,
> next to the progress indicator, but you are welcome to do something else.
>  - If I record and then type a link into the address field the page comes
> up but is not added to the test tree.
>  - You may want to allow the delay between user clicks while recording to
> be stored and used as the pre-load delay.  Users should be able to enable
> and disable this feature.  The pre-load feature allows the playback to
> occur in a more realistic fashion (i.e. simulate user delay).

Will do.

> All in all though I really like the recording.  I am glad someone did it.
>
> :-)  I am curious about your "added support for variables" comment in your
>
> CVS entry...what exactly is it?

At this stage variables are only supported in URLs.  I needed them because 
the site that I am testing uses a system assigned session ID in the URL 
(rather than cookies).  When I test my site the login page is loaded and this 
has a session ID as part of the login form's action URL.  I use a "Regex 
Variable Reader" to read a value for the variable SESS from the page using a 
regular expression.

In the test properties dialog I set the session substitution command to 
replace any session IDs in recorded URLs with $SESS$.  Hence when the 
recorder adds resources they have $SESS$ instead of the session ID.  When the 
browser opens these resources it substitutes $SESS$ with the value read by 
the variable reader on the login page.

While variables are only substituted in URLs at this stage I imagine it could 
be useful to allow them to be substituted in request parameters (i.e. post 
data) as well.  This means you could read a variable of one page and then 
submit it as part of the post request on another page.

> BTW - Thanks for your hard work and input on LogiTest.
>
> -Anthony

No problems - likewise for yours.

Clancy.
__________________________________________________________________________
Clancy Malcolm <cl...@cy...>       Application Development Manager
Cybersource: Level 9, 140 Queen Street, Melbourne  http://www.cyber.com.au
Phone: 03 9642 5997           Fax: 03 9642 5998        Mobile: 0414 297879