RFC 44 Complete

Configure Direction

Andi, now that I can find and parse buildconf.xml, locale.php.xml, and
package.xml files what do I do?

I need some direction.

How do I transform a package.xml file? I can parse it with PackageParser and
then what?

Charlie

E-Mail

Re: Tries

Charlie,
    i had a go at this again this weekend, and i came up with a brainfreeze
that i think you can resolve.
    the single greatest issue i have with my code is that to be general
enough to be included in PHP itself, it cannot special-case wildcards. this
means essentially that all paths containing wildcards would have to use you
code as it exists. one  i thought of was to check normally. if the key does
not exist, use the current trie code. this would only be worthwile if the
bulk of the paths did not contain wildcards.

in any case, in looking at PHP's code, i can probably get away with
modifying array_key_exists to add the new functionality. IOW, if the key is
a string or long, it functions as is. if an array is passed, we have our
trie search functionality. i think this is sufficiently intuitive and useful
to pass muster as a patch to the PHP source.

_cc_


----- Original Message -----
From: "Charlie Killian" <Charlie@...>
To: <dev@...>
Sent: Thursday, March 28, 2002 2:02 PM
Subject: RE: [binarycloud-dev] Tries

> >     i had a bit of idle time this morning so i decided to experiment
with
> > the path access problem (accessing a value at an arbitrary index in a
> > multidim array). a solution was fairly simple to code, though ill have
to
> > wait till tonight (EST) to run some tests.
>
> Alright! Let us know how the tests turn out.
>
> Charlie
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@...
> For additional commands, e-mail: dev-help@...

RE: Tries

> one  i thought of was to check normally. if
> the key does
> not exist, use the current trie code. this would only be worthwile if the
> bulk of the paths did not contain wildcards.

Not being able to check wildcards is okay. Most of the look ups we'll be
dealing with are paths without wildcards and I think there will be a
significant time savings performing the look up in C instead of PHP.

> in any case, in looking at PHP's code, i can probably get away with
> modifying array_key_exists to add the new functionality. IOW, if
> the key is
> a string or long, it functions as is. if an array is passed, we have our
> trie search functionality.

This is a good method. I like it.

> i think this is sufficiently intuitive
> and useful
> to pass muster as a patch to the PHP source.

It is intuitive and useful. Let me know when the patch is ready.

Also, can you fix array_merge_recursive()? It currently doesn't clobber the
value if it already exists, which is stupid. Take a look at the user
comments on:
http://www.php.net/manual/en/function.array-merge-recursive.php

Thanks, for helping out on array_key_exists(). It will make a big time
difference in the BC takes to execute.

Charlie

Re: Tries

Charlie
> This is a good method. I like it.
>
> > i think this is sufficiently intuitive
> > and useful
> > to pass muster as a patch to the PHP source.
>
> It is intuitive and useful. Let me know when the patch is ready.
i can probably send you something in a day or so. im not working from CVS so
ill just send you some code you can substitute for the current
array_key_exists().
    btw - is there still a need for array_get_value() ? i can think of a few
uses. the code base is exactly the same.

> Also, can you fix array_merge_recursive()? It currently doesn't clobber
the
> value if it already exists, which is stupid. Take a look at the user
> comments on:
> http://www.php.net/manual/en/function.array-merge-recursive.php
>
i'll have a look. has anyone filed a bug report/feature request on the issue
yet ?

-cc-

Re: UML

Hi Alex,

> anyway: have a look, there are mini-tweaks to some things,
> and some not-so-mini tweaks to others. (attached gif)

Looks very good.

> I'm not done yet, I don't like Container extends Widget...
> and I'm considering a Module factory.

Sounds interesting. I think it makes sense.

> Also note the binarycloud class, (init with more methods,
> basically) that will replace global objects. So there will
> be a single binarycloud object in the global namespace, no
> other pollution. I like that a lot.

Me too.

> Debug's name is Log in this model, I'd like people's
> opinions on that.

Agreed. Log is the basic IO class for messages. Debug are just the modules 
that utilize log for a special type of messages. Like that a lot.

> Log will also be extended (and configured) optionally to
> push BC events to syslog,

Groovy !!!!!!!

> --------------
> Andi: what about hooking up request to session
> vars? Does it do that now?
> --------------

You mean retrieving sess vars? That's alreay implemented! It does handle 
all kind of vars from the outside, session too. But it does not modify 
session vars. I think it should not do this, because it's the job of 
Sess.php.

> to see what everyone thinks, get the stuff done, and as
> Phing2 finishes off get all the widgets and a functioning
> entity pipeline going.

I'm all for it.

> Then we can release.

Harr harr harr, the magic words!

Andi

Metabase QueryAll()

Does anybody know if Metabase's QueryAll() method can return results in an
associative array?

So a call to $_db->QueryAll($query, $result_array, $types);

Would return:
[0] => Array
        (
            [id] => 1
            [desc] => Sally's really long name.
        )

Instead of:
[0] => Array
        (
            [0] => 1
            [1] => Sally's really long name.
        )

Charlie

Re: #binarycloud on OpenProjectsNetwork

Hi Dave,

> I got #binarycloud setup on the open projects network today.
> irc.openprojects.net // #binarycloud

I always get kicked due some ident errors. My firewall is blocking ident 
requests. Is that an issue with the irc server?

Andi

Re: Commit Trie Class

Hi Charlie,

> I've commit the Trie class to ext/phing/classes/phing/core/
> and an example script to ext/phing/tests/ .

Fantastic.

> The Trie class operates on arrays of values. So to set a
> node to TRUE pass  array("path", "to", "node") and TRUE to
> $Trie->SetNode().

Cool concept. How about using the dot-notation there additionally:

SetNode("path.to.node", true);

So method check if string, if yes $path = explode(".", $string);

> The public methods of the Trie class are:
> GetTrie()
> SetNode()
> GetNode()
> IsNode()
> RemoveNode()
> Clear()

Can we pass something to GetTrie() to get a subtrie, i.e.

GetTrie("path.to")

> I'll work on some Trie Utilities next. Some utilities that
> convert a file of paths to a Trie object.

Very Cool.

> To give an idea of the Trie class below is the output from
> the example

Fantastic work Charlie.

BTW: Happy easter to all of you!

Andi