Welcome to the second edition of the CakePHP weekly summaries.
Its quite embarrassing that it had to take so long time to get the second editon out. Too long time has passed and the amount of things that has happend during that period is waaay too big to be covered here.
Therefor I present this half-done 2nd edition of the weekly summaries, with a promise that every week from now on I will give you the latest and greatest from the CakePHP world.
So with no further ado, please enjoy this old and probably rather outdated edition of the weeklymonthly summaries. See you at the 3rd edition soon

Trac changes

• Changeset #6321 (phpnut) Fixes Ticket #3817, Some non-executable files in source tree have executable permissions
• Changeset #6322 (phpnut) Updating controller tests created with the console. Added associated fixtures to model tests
• Changeset #6323 (gwoo) Updating Cache to apply prefix for every key, closes Ticket #3821
• Changeset #6324 (phpnut) Correcting links generated to additional app tests
• Changeset #6325 (phpnut) Adding support for test fixture generation when creating models, this will eventually be moved to a task. Currently the $records are hard coded to create field type specific data, this will eventually be changed to create random data based on the field type. Refactored tests that are created for models, 3 working tests are not created when generating tests for the model. Added notice output to terminal when creating a project, the CAKE_CORE_INCLUDE_PATH is automatically altered and may need to be changed when moving to a production environment. ”
• Changeset #6326 (phpnut) Fixes svn file property line
• Changeset #6327 (phpnut) Removing unneeded ‘ : ‘
• Changeset #6328 (phpnut) Removing unneeded ‘ : ‘
• Changeset #6329 (nate) Fixing reading and writing in Configure class for references 3 levels deep, fixes Ticket #3845
• Changeset #6330 (nate) Removing un-needed method from TimeHelper? test
• Changeset #6331 (phpnut) Removing var $useTable; from generated model when tables follow conventions Fixed issue with model tests and fixtures not being created properly when Model name does not match table name
• Changeset #6332 (phpnut) Updating controller tests created with the console. Added associated fixtures to model tests
• Changeset #6333 (phpnut) Fixes Ticket #3849, Deprecated loadView() call in email component when using theme
• Changeset #6334 (gwoo) updating HttpSocket? with auth, closes Ticket #3838 removing type casting, fixes Ticket #3827 adding socket.group to tests
• Changeset #6335 (gwoo) updating component, removing use of Router not needed anymore, fixes Ticket #3858
• Changeset #6336 (nate) Adding non-null association key fix in Model, and updating Model::isUnique() to account for current record, fixes [ticket fixed 38599, closes [ticket fixed #2032]
• Changeset #6337 (nate) Fixing return value in Model::saveAll(), fixes Ticket #3852
• Changeset #6338 (phpnut) Fixes Ticket #3829, Scaffolding and bake script has problems with tables that has underscore in it’s name
• Changeset #6339 (phpnut) Fixes Ticket #3798, custom scaffolding file lookup paths inconsistent
• Changeset #6340 (phpnut) Fixes Ticket #3862, set_time_limit(0) has no effect because of ini_set(‘max_execution_time’, 60 * 5); changed max_execution_time , 0. Since php 4.3 this is the default anyway
• Changeset #6341 (phpnut) Fixes Ticket #3811, $_SESSION not found Fixes [ticked fixed 3823], Session not restarted after instantiating new SessionComponent? Fixes #3850, Configuration Session.start and SessionComponent::__active aren’t handle correctly
• Changeset #6342 (phpnut) Fixes Ticket #3851, Acl component fails to check() if a permission is denied in action (*)
• Changeset #6343 (phpnut) Adding test cases from [ticked closed 3810]
• Changeset #6344 (phpnut) Reverting [] and [] this commit broke the working implementation. Closes Ticket #3810
• Changeset #6345 (phpnut) Fixes Ticket #3854, Scaffolding is generating an id field with an empty value for auto_increment primary key
• Changeset #6346 (phpnut) Fixing bug with scaffold and valid Session id check
• Changeset #6347 (phpnut) Fixing bug with scaffold and valid Session id check in 1.1.x.x
• Changeset #6348 (gwoo) updating View error handling for extensions, fixes Ticket #3795
• Changeset #6349 (nate) Correcting parameter declaration in API shell
• Changeset #6350 (nate) Fixing database drivers for UPDATE/DELETE changes, updating drivers to comply with model tests, most passing. Fixes Ticket #3794
• Changeset #6351 (nate) Whitespace fixes
• Changeset #6352 (gwoo) adding connection param to schema shell
• Changeset #6353 (nate) Enabling alternate titles for PaginatorHelper::sort() links based on sort direction, closes Ticket #3865
• Changeset #6354 (nate) Removing MySQL-specific schema code from core, improving support for PostgreSQL sequences, most tests passing

Google group topics

Why doesn't Find* return a group of model objects?

This topic has been discussed many times before, but Deane felt that it was required to ask yet again: “Why doesn't Find* return a group of model objects?“. He wanted to be able to work with his models in the view (!BAD DESIGN!), and was confused why he only got an array back from the find* methods. Gwoo replied, and I quote: “Speed, simplicity, clarity” and pointed out that Cake provide view helpers to work with the arrays in almost any way you can imagine, and that the models provide callback functions (afterFind, afterSave, beforeFind, beforeFind) to format your data.
Deane thanked gwoo for the reply, but “We couldn’t disagree more strenuously, and we think you’re wrong beyond belief in your reasoning, but we appreciate the response”…. didnt like gwoo’s reply at all.
Gwoo replied in a calmly matter that “Everyone is entitled to their opinion and their choice of frameworks.”
rtconner joined the discussion, offering a solution based on some old code by gwoo that would enable the models to return objects (StdClass, not models classes)instead of arrays.
Nick also joined the thread, backing up Deaen’s point of view (Cake is wrong, symfony lover is right) that its proper OO to use objects instead of arrays for the view data, but also reasoned that cake does indeed have the needed hooks and concepts to achieve Deane’s initial idea with a minimal effort. He explained that in a project he had created with CakePHP was that relationships (belongsTo, hasMany ect.) get more complicated with the OO styled approach, since the queries often fetched much more data than was really needed in the view. The combination of the controller didnt know wich fields the view required and the complex relation ships was in his opinion, not ideal at all. Their work a round was to create a “LazyData” class to handle it.
Robby Anderson dropped in his 2 cents, and agreed that the cake was isnt 100% pure OO, but that the concept of “having the view incapable of performing any meaningful business logic” is bad, and will sooner or later lead to an unmanagelbe code base.
Deane, the creater of the thread, posted a public apoligy to gwoo, explaining that he was fustrated at the time when he posted his previous post.

So, the summary must be that while some people dont like the design decission that CakePHP models return arrays instead of objects, others agree that its the right way to encapsulate the buisness logic in a correct Model View Control manner, and if you want to break theese convensions, its rather easy to do with minimal effort.

Passing Data To A Form

Travis asked how you are Passing Data To A Form. A bit unexpted (or perhaps expected given the flood of duplicate / newbie questions hitting the groups daily) nobody answered his question. The answer is pretty simple, and covered in any tutorial aviable on cakephp website AND in the documenation / manual.

1
2
3

function update($employeeID)  { 
     $this->data =  $this->Employee->findAllById($employeeID); 
}

The Show Is Back!

Jeff Loiselle aka phishy announced that the show is back. No replies, no thing, its sad… the show rocks !

I18n in 1.2 model validation message

Athies asked how does i18n in 1.2 model validation message work. His iniital attemp was to use the i18n method ‘__()’ in the model $validate array, but he was getting syntax errors when doing that.
Since PHP does not support useage of functions in instance delclarations its not possible to use i18n directly in the validate array.
Francky06l pointed at his solution in another google post i18n: static language references with __("My lang key", true) in models and controllers, and while the solution will work, its far from ideal when working with i18n.
A CakePHP google group regular, Dr. Tarique Sani, pointed out that the correct way to handle i18n validation messages was in the view, not at the model level. Amit Badkas backed up Tarique up and provided a simple example on how to do it:

1
2
3
4
5

< ?php echo 
$form->input('username', array('error' => array( 
    'alphanumeric' => __('Username must contain alpha-numeric characters', true), 
    'between' => __('Username must be between 8 to 20 characters', true)))); 
?>

Francky06l seemed to know that, his consern was the if he multiple views where the validation errors should show, it would be convenient to do it at model level.
Robby Anderson wasnt sure why you just couldnt do both, create the validation errors in the model, and wrap them in __() in the view.

And sure, that will work – but – you will loose some cake magic with that. The console application

cake i18n extract

will not add the i18n messages to the pot file automagic. Also, the i18n extract command will not even search your models for any i18n magic. It will only look in controllers and views. You can choose to call this a limit or a feature, but thats how cakephp has decided to do it, to enforce their paradime on how an application should be build and structured.

FormHelper Labels

Travis asked how it was possible to disable the automagic label in the FormHelper. Robby Anderson provided the correct solution for this issue, use ‘label’ => false:

< ?php echo $form->input('Employee.name',array('label' => false));?>

Using model objects in the controllers

Deane from earlier posted another question on Using model objects in the controllers. I dont really want to dwell much on this post, since its really attemp to take CakePHP in another direction than its supposed to. Deane clearly shows a lack of insight on how cakephp is supposed to work, and failed to provide any usefull examples that could show the group what he wanted to do.

Sorry that I havent created the weekly summaries for almost 1.5 week now. The reason is that I have begun on a new job, and its consuming quick alot of time and energy Lot of new systems to learn (They dont use Cake…. yet, booooh!)

I have the second edition 80% done, just need to clean it up and proof read it a few times Should be done before this weekend

This week alot happened, but the biggest news is that CakePHP 1.2 have entered BETA phase, and with that – we are close to a feature (API) freeze of the 1.2 branch.

•••••••••••••••••••••••••••••••••••••••••••••••••••••••

CakePHP 1.2 beta

Yes, finally it happened. Please look at my post about the .

Trac changes

• Changeset #6288 (gwoo) Added support for stored procedues in DboMysqli, fixes Ticket #2340
• Changeset #6289 (gwoo) Updating controller task with i18n wrappers, closes Ticket #3490
• Changeset #6290 (nate) Updating help text for ACL shell interface, fixes Ticket #3790
• Changeset #6291 (phpnut) Moved merging of components, helpers, and uses vars to Controller::_mergeVars(); fixes bug with PluginAppController? vars not being merged before components are loaded. Fixed Ticket #3760
• Changeset #6292 (phpnut) Missing keys when Model merges actsAs. Fixed Ticket #3786
• Changeset #6293 (nate) Adding quoting and name-resolving fields in TreeBehavior. Fixed Ticket #3778
• Changeset #6295 A few new exciting features was pushed into the repository:
  • Model::generateList() was deprecated, and replaced by the new 1.2 cake find syntac. Model::find(‘list’)
  • Model::saveAll() was enhanced to save multiple records and associated records
  • Some transactional methods (begin, commit, rollback) was added to DataSource class as well
  • bindModel now support to change the bind settings on the fly if the model has already been bound
  • Fixed Ticket #3720, Ticket #2355, Ticket #3615
• Changeset #6296 Implemented Validation::extension(); and Vaidation::range(); Fixed Ticket #2897
• Changeset #6297 Some optimization in code. Fixed Ticket #3653
• Changeset #6298 Adding test for root path caching. Fixed Ticket #1349
• Changeset #6299 Adding full join model support. Fixed Ticket #828, Ticket #1045
• Changeset #6300 Adding Debugger::log and test cases. Fixed [ticket tixed 995]
• Changeset #6301 CSRF usability problems, Security Component and multiple instances of the webapp. Fixed Ticket #2608, Ticket #3436
• Changeset #6302 Adding new URL to validation test
• Changeset #6303 Renaming domains in test to generic names
• Changeset #6304 Fixing Model::find(‘list’) when using self-joins
• Changeset #6306 Refactoring, fixing SessionComponent test, adding String::tokenize()
• Changeset #6307 Adding regular expression support to Set::extract() [ Set::map($data, '{s}.{n}.{[0-9]} ]. Fixed Ticket #3263
• Changeset #6308 Fixing paths returned by Debugger::trimPath();
• Changeset #6309 Fixing paths in debugger tests
• Changeset #6310 Fixing paths in debugger tests
• Changeset #6312 Adding String class to shell boostrap. Fixed Ticket #3793
• Changeset #6313 Fixing JavascriptHelper::link() for script files outside of webroot/js (i.e. plugins). Fixed Ticket #3799
• Changeset #6314 Removing unused directories from app/pages. Merging changes from app/ to skel/
• Changeset #6315 Fixing after filters for Model::find(‘list’). Fixed Ticket #3805
• Changeset #6316 Updating TimeHelper. Fixed Ticket #3801
• Changeset #6317 Bug in hasAndBelongsToMany. Fixed Ticket #3814
• Changeset #6320 updating Set::map() handling of deeply nested arrays, thanks to speedmax for an updated test case. Fixed Ticket #3809

Google group topics

How to find referrer url?

A guy called ‘krr’ opened a question how to find referrer url?, and Daniel Hofstetter quickly replied that Controller::referer() would return the referer.

Will CakePHP be able to survive as the Zend Framework matures?

The framework war ignited yet again when “Action” created Will CakePHP be able to survive as the Zend Framework matures?.

Upgrade problems

Following the release of 1.2 beta “Mech7″ Upgrade problems after upgrading his application to the new beta verion. Chris Hartjes quite replied that Mech7 should check his apache and php error logs. Mech7 however, quickly replied that the error was solved (after only 9 minutes), and his solution was to clear the app/tmp/cache folder in this app.
I had however already written a list of steps I follow when my application suffers from the White Screen Of Death:

• Check that debug > 0
• Clear app/tmp/cache/persistent/*
• Check local vhost error log
• Check global httpd-error log
• Restart apache, (stop, start), restart cache services (memcached ect.)
• Enter the world of die(‘stop’); from app/webroot/index.php and downward in your application

Table inheritance, multiple schemas, and associations

Brian had a interesting question, well, for us postgresql users at least, table inheritance, multiple schemas, and associations. There we no response to this thread, but after a few hours, he returned with a working solution, very interesting stuff.

Depricated ListControllers

Ronald Chaplin asked what the ‘new’ way was in 1.2 to get a list of controllers in your application, since 1.2 had Depricated ListControllers ( listClasses( CONTROLLERS ); ). He had tried Configure::listObjects(‘controller’, ‘/app/controller’);, but without success.
Arno Esterhuizen quickly pointed Ronald in the right direction, the CakePHP 1.2 api

OthAuth looses session – upgrading to new 1.2.0.6311 (2008)

Klaus had a problem, othAuth looses session – upgrading to new 1.2.0.6311 (2008), but so far no replies, might be a local issue on his setup?

Themes management for an app ?

Max asked if anyone had a component/helper for Themes management for an app ?. Tarique shamelessly pointed to his own blog, where there is quick guide on how to make cakephp use ThemeView.

Cakephp with PHP 5 and Mysql 5

One of the too often asked questions on the group was asked yet again, Cakephp with PHP 5 and Mysql 5, and yes. CakePHP works fine with both MySQL5 and PHP5!!. Its rather funny, or sad depending on who you are, that the same questions is asked over and over on the googlegroups…. I’m sure the CakePHP guys chose google for a reason

Bug for find('All'

Another of repeating posts on the mailing list is the “I think I have found a bug” or just bug for find('All' in this case. And as always, people always point directly to Trac bugtracket, as its the one and only place to submit possible bugs..

CakeFest website Launched

Mariano Iglesias announced that the CakeFest website Launched! There was no replies or reactions on the list, I just hope that people did not overlook this little gem

Bakery hates me

Adam Royle claimed that “bakery hates me“, because he was not able to comment on a bakery article. Mariano Iglesias was quick to point to tracwebsite, but a Dr. Hannibal Lecter was snappy, and said “I don’t think posting links to Trac solves anything, it is a fact that bakery has many issues (some minor and some..not so minor). Some things should be seriously reconsidered and rewritten. But that’s just my 2c.”.

As much as Chris Hartjes agreed with Dr. Hannibal Lecter, he pointed out the important fact that trac is still the one and only place for bugs on cakephp related projects.
Dr. Hannibal Lecter hoped that someone from the core team was following the group, since it wouldn’t be very (And I quote) “userfriendly community” if they didnt.
Dr. Tarique Sani (Whats up with all those doctors?) replied that it was unfair against the php team to mark the community less userfriendly because users finding bugs wouldn’t bother to create tickets on trac, and just posting the errors on the mailing list.
And FWIW my personal experience is that posting in trac works

Adam Royle said that the reason why he hadn’t posted a bug report on trac yet was because he didn’t like opening tickets without a patch or solid way to reproduce the error.
Dr. Hannibal Lecter returned a apologized that it was not his intention to be unfair against the CakePHP team, and that he really appreciated their efforts. He was however puzzled that the creators of such a great framework as CakePHP would neglect The Bakeryand made a rather good point: If new users use the bakery, and its full of errors, what would they think of the underlying framework as well?
Adam Royle returned to the discussion and tried to get answers on his two original questions

• He was unhappy wit the quality of this article on the bakery
• What version of cake is the bakery running on? (1.2 pre-historic afaik)

Chris Hartjes, our favorite grumpy Canadian, closed the thread with a few harsh comments I’m not going to quote here Larry E. Masters aka PhpNut posted his announcement on the new 1.2 beta at: New Year, New Beta, that received a few “thanks”, but nothing else of interested.

Require SSL for entire site?

Bryan Encina asked if it was possible to Require SSL for entire site?. His own idea was to check $_SERVER['HTTPS'] in his AppController, and then redirect based on this. Chris Hartjes questioned the safety of $_SERVER['HTTPS'] (its perfectly safe), to his understanding, it was possible to spoof (its not) this variable. His suggestion was to limit access on webserver level instead of application level.

I (Christian Winther) pointed out that mod_rewrite would be a perfect solution for this problem, since the RewriteCond directive could check for both SERVER_PORT (port 80 = non ssl, port 443 = ssl) and SERVER_PROTOCOL (http = non ssl, https = ssl)
. Robby Anderson suggested using the SecurityComponent::requireSecure() method instead, also a possible solution if you wish to let ssl management be up to the application (htaccess will most likely perform alot better if its just for redirect).
Chris Hartjes returned to the thread because he had felt a bit wierd about his earlier reply to the group, and confirmed that $_SERVER['HTTPS'] indeed cannot be spoofed. He suggested using Robby’s solution.