Selenium would help immensely. During the 2.8 coding cycle, I
refactored the web unit tests so that they would run without any
external dependencies. The rationale was that we needed to make it
easier to create and run web unit tests. Certainly, we can (and
should) use Selenium more to test JavaScript functionality.
The nice thing about Selenium is that you can use the Firefox Selenium-
IDE plugin to "record" the test session. Then, with a little light
editing you can turn it into test that runs as part of the "webtests"
Ant target. See the mostly-accurate docs in tests/etc/selenium/
readme.txt.
The bad thing about Selenium is that it does NOT seem to work with
Firefox 3 yet. However, Firefox 2, IE, Safari etc. seem to work fine.
Regarding the other comments about JavaScript dependencies in JSPWiki,
everyone has made good points so far. I agree that we will need move
certain functions from the client to the server at some point --
ESPECIALLY for any client-side code that manipulates session state. By
"manipulating session state," I MOSTLY mean server-side (JSP, servlet,
core) code whose processing depends on client-supplied cookies.
In the short term, we should probably file JIRA issues as reminders
for things that should be moved.
Andrew
On Aug 20, 2008, at 12:15 PM, Harry Metske wrote:
The point is that we don't do enough automated testing, it would be
nice if
we had some kind of JUnit tests for JS.
Can the selenium tests do something here ?
What is the "industry standard" for JS testing ?
I, also like to keep all the goodies of JS in JSPWiki on board, and
let's
get to the Stripes release as soon as we can.
regards,
Harry
2008/8/20 Janne Jalkanen <[EMAIL PROTECTED]>
On Aug 20, 2008, at 04:42 , Terry Steichen wrote:
I've been wondering for some time if JSPWiki hasn't become somewhat
too
dependent on esoteric Javascript code for its core functionality.
Dirk
has done a wonderful job - of that, there's absolutely no doubt.
When
the template logic works (as it normally does quite well), it's a
marvel
to behold.
But when it doesn't, it's sometimes a nightmare to figure out.
I'm OK
figuring out Java, servlets, HTML and the like - even with its poor
documentation, I can eventually figure out the innards of JSPWiki
code.
But the increasingly complex Javascript will often does me in.
I think the problem partly stems from the fact that Dirk is the
only one
doing the JS hacking. With the Java code, it's a shared effort by
many
people, so it tends to stay readable across for least a few people,
which
makes it readable to the rest as well.
I think we need a few more Javascript hackers ;-)
(And also we'll need to think which of the JS functionalities would
be
better done server-side. For example, I think section editing
would be
nicer to do server-side, not client-side - for the simple reason
that then
we could also lock parts of the page, or also be completely
collaborative.)
/Janne
--
met vriendelijke groet,
Harry Metske
Telnr. +31-548-512395
Mobile +31-6-51898081
