John Kleijn - Comments

John Kleijn: ExtJS RecordFormPanel

nospam@example.com (John Kleijn) — Wed, 29 Dec 2010 21:50:48 +0100

That makes sense, I didn't know StoreMgr.lookup() acts as a factory. It's not something I would personally use, as you point out the primary function of this component is a "shared store" situation. It might be useful in situations where you need a local store though, but then again, that is not a logical use case for this component.

For sake of completeness it's a good suggestion though

Mchl: ExtJS RecordFormPanel

nospam@example.com (Mchl) — Wed, 29 Dec 2010 21:38:22 +0100

It's not for referencing by ID (though Ext.StoreMgr.lookup()'s signature might suggest that). It's for being able to do something like this.

var recordForm = {
xtype: 'recordform',
...
store: {
xtype: 'jsonstore',
...
}
}

In other way it lets you defer instantiation of the store until you actually need it.

Granted, this component will probably be used together with other store enabled components like GridPanel, and would be using stores already associated with them (via Ext.grid.GridPanel.getStore() for example). Nonetheless, adding this one line adds some extra functionality to the RecordPanel which makes it a bit more versatile.

John Kleijn: ExtJS RecordFormPanel

nospam@example.com (John Kleijn) — Wed, 29 Dec 2010 21:14:30 +0100

I guess. Personally I am not really using referencing by ID, I prefer a stateful interface. I use it for combo's though, for often referenced lists, like an employee list. Technically I should not do that though, for the same reasons a Singleton Registry is ill-advised.

Mchl: ExtJS RecordFormPanel

nospam@example.com (Mchl) — Wed, 29 Dec 2010 21:04:16 +0100

You might want to add

this.store = Ext.StoreMgr.lookup(this.store);

to initComponent() function. This will allow passing both config objects as well as Store objects. See source of Ext.grid.GridPanel.initComponent()

John Kleijn: Topological sorting of JavaScript files

nospam@example.com (John Kleijn) — Wed, 29 Dec 2010 21:01:58 +0100

Not really. I don't scan for xtypes myself (I use a @depends JSDoc tag), but adapting the code above to work for xtypes is trivial.

Just add a sibling to the 'classes' index in the spec, eg:

preg_match_all("Ext.reg(ister)?\(('|\")([a-z]+)('|\")", $contents, $matches);

'xtype' => $matches[3]

Then simply add another iteration for every file, if a match for "xtype:\s*('|\")" . $specA['xtype']. "('|\")" is found, add the link.

PS Serendipity sucks ass.

Mchl: Topological sorting of JavaScript files

nospam@example.com (Mchl) — Wed, 29 Dec 2010 20:44:01 +0100

My solution was a bit uglier. I put every component into a separate file and add a list of it's direct dependencies (ugh!). Then I let PHP recursively load all JS files needed for any given 'Screen' (a basic unit of my GUI system).

I'm currently thinking on how to remove dependency lists by parsing the source code for used classes (might be tricky when lazy instantiation through xtype is used).

Mchl: Why you can't (or shouldn't) unserialize exceptions

nospam@example.com (Mchl) — Fri, 24 Dec 2010 00:40:27 +0100

Hi John,

This is something I wouldn't ever think of. Lesson for today: (if it's not necessary) build your classes in such a way, they won't throw Exceptions on __wakeup()

Anonymous: Story: Doctrine 2 lead up

nospam@example.com () — Mon, 26 Jul 2010 15:30:43 +0200

Hey Ross,

I don't think it's ready for a public release. It's very basic anyway: just a helper to do insert node, remove branch, fetch branch and move branch.

It's not integrated into the persisters or anything. Hierarchical objects need to extend a base class that handles the left, right and level values (using @MappedSuperclass).

There are definite downsides to this approach, in perticular the list of children will not always be populated, only if fetched using the helper.

Add to that that technically I do not own this code, I'll have to decline.

Ross: Story: Doctrine 2 lead up

nospam@example.com (Ross) — Mon, 26 Jul 2010 14:53:50 +0200

On the subject of Doctrine2, any interest in opening up your NestedSet implementation?

David Weinraub: NetBeans is TDD unfriendly!

nospam@example.com (David Weinraub) — Wed, 19 May 2010 12:37:39 +0200

Right, I'm on WinXP, so it's probably the Java slowing up the works. My PHP unit tests run significantly faster from command line (no surprise). Cheers!

John Kleijn: NetBeans is TDD unfriendly!

nospam@example.com (John Kleijn) — Wed, 19 May 2010 12:07:18 +0200

No problem. It seems though as if my verdict of Zend Studio was premature. I'm starting to have similar issues with NetBeans, which probably means the issue with Java on Ubuntu, rather than with either IDEs.

David Weinraub: NetBeans is TDD unfriendly!

nospam@example.com (David Weinraub) — Fri, 14 May 2010 08:55:32 +0200

Thanks for your post. I was wondering why NB always failed to find my tests. As you explain, the test for MyClass must also be named MyClass, but located within the /tests folder. Like you, I prefer to have my test class named MyClassTest, though I am perfectly happy to place it in a separate hierarchy in the /tests folder.

I confess that I am not doing much pure TDD, so I usually do have my SUT class already defined, if not completely written. So, as you observe, I can run tests by doing a right-click Run on the Test file itself, provided NB is pointed to the right test bootstrap file.

Now to figure out why my tests run so slowly, even via this method.

Thanks and cheers!

ficuscr: Zend Framework 1.7: Zend_Db_Table still sucks

nospam@example.com (ficuscr) — Wed, 07 Apr 2010 23:39:21 +0200

Amen!

Mchl: Google not indexing my blog

nospam@example.com (Mchl) — Wed, 07 Apr 2010 22:08:39 +0200

'John Klein' folder in my Thunderbird just lit up blue and indicates there's a lot of new items in it... I guess it means things are getting back on track.

John Kleijn: Started on a new framework

nospam@example.com (John Kleijn) — Sun, 28 Mar 2010 03:47:43 +0200

Hi Daniel,

Yes, I do mention that ("arguably we could already do that by recursively casting to an array").

Thanks for pointing out the mistake in the text, pretty sloppy of me.