Hi all -- with the most recent checkins, I knocked #5 and #6 off the
list. So, the WikiBackgroundThreads are now JMX-triggered classes. And
StripesURLConstructor is now the default (and only allowed)
URLConstructor. Amazingly, after I eliminated one dumb bug, it passed
198 out of 200 JSPWikiMarkupParser tests. (Two pesky Scandic tests
aren't passing...)
I'll focus next on eliminating the 55 or so failing tests, and
possibly the JSON death issue.
Andrew
On Feb 7, 2010, at 16:58, Dirk Frederickx <[email protected]>
wrote:
The improvements on the .jsp's are cool, and really a simplification !
Stripes is really a big improv.
I just need to find the time to catch up with most of your changes.
In parallel I'm making some js enhancements, especially on the
editor front.
But currently I have no working environment, so it's hard to start
tracing
and fixing stuff.
dirk
On Sun, Feb 7, 2010 at 10:41 PM, Andrew Jaquith
<[email protected]>wrote:
Thanks, Dirk!
I understand what you mean about the changes to the templates But
the
templates are finally done. I don't expect them to change much
more. The
most difficult work is behind us.
If ypu think it would speed things along, we could have a quick call
(iChat?) about the Javascript issues.
Andrew
On Feb 7, 2010, at 13:12, Dirk Frederickx <[email protected]>
wrote:
Andrew,
A bunch of javascript updates are waiting for check-in.
The speed at which the template is changing is hard to follow ;-)
Currently, I'm blocked with deploy issues -- hope this can be
fixed soon.
JSPWIKI-382 can be closed soon: posteditor.js is already not used
anymore;
but the new plain editor is still buggy.
dirk
On Sat, Feb 6, 2010 at 4:36 PM, Andrew Jaquith
<[email protected]>wrote:
Harry, thanks for the kind words!
The first two issues are good ones that we should fix. +1. Good
catch!
Note
that the JSON search needs to be rewired to use the Stripes calls.
Ultimately we want to get rid of the JSON bridge and just call
the event
URLs directly.
Not sure we need to do web unit tests just yet. Definitely for beta
though!
Andrew
On Feb 5, 2010, at 9:30, Harry Metske <[email protected]>
wrote:
Andrew,
first I want to express my appreciation for the huge
contribution you've
made, we should have more people on the team like you.
Now the issues, I can agree on all 6 issues, and want to suggest
we add
the
following to them assuming we want an Alpha JSPWiki that works
for basic
functions like page editing, saving, and so on.
- There is still the "page edit concatenation" problem (no JIRA
issue
yet,
but I can create one)
- JSPWIKI-510 - SearchManager.JSONSearch.findPages() does not
honor ACLs
(security is important for the ASF and for ourselves)
- not quite sure if we should also first try to get the webtests
(Selenium
or an alternative?) up and running
One of the things I can help with is fixing the unit tests I
think.
regards,
Harry
2010/2/4 Andrew Jaquith <[email protected]>
Hi everybody,
If you've been paying close attention to my recent commits ---
and I
*know* you have :) --- you know that I've been putting a lot of
energy
around re-working the JSPs. As of last night's checkins, I feel
that
most of the goals I set for the 3.0 view layer have been
achieved.
Scriptlet code has been moved into the ActionBeans, markup has
been
"modernized" with JSTL, and the top-level JSPs have been folded
into
the template JSPs. The net result is a set of JSPs that are
slimmer,
simpler and better organized. A few stray bugs (notably:
JavaScript
and a few URLBinding issues) remain, but the hard work is done.
It is time to turn our sights towards getting an Alpha build
out the
door. Here are the current blockers as listed in JIRA:
1. JSPWIKI-303 JSPWiki-API library creation.
2. JSPWIKI-421 JCR backend
3. JSPWIKI-382 Remove posteditor.js
We should make some decisions soon about whether these are really
blockers or not, and figure out what we need to do to close
them if
they are.
I'd also propose solving three more issues before we can
declare Alpha:
4. Clean unit tests. We still have about 30 renaming,
ReferenceManager
and related plugin tests that are failing. I *think* they might
be
related to a recent Priha vintage. The build should run clean
before
we release an Alpha.
5. URLConstructors. With the physical top-level JSPs gone, we
now rely
on Stripes URLBindings to map incoming *.jsp resquests to
ActionBean
event. We should do the same for outgoing URL **generation** by
making
StripesURLConstructor the default. While we're at it, we should
kill
the other URLConstructors, because FileBasedActionResolver will
allow
URLBindings to be externally defined. For backwards
compatibility,
ShortUrlRedirectFilter allows legacy short URLs to be safely
intercepted and redirected.
6. Refactor WikiBackgroundThread abstract class as JMX timer
MBeans.
This would eliminate our somewhat unreliable timer
implementation.
Today, background threads don't kill themselves reliably. I
take full
responsibility for this, but I also can't fix it easily, and
would
rather do it through JMX.
I can do 5 and 6 fairly quickly if we agree to do them.
Are these good priorities for Alpha? What else are we missing?
Andrew
