Comments for Davey Shafik

Comment on Mimetypes (and APIs) by John Congdon

John Congdon — Wed, 25 Apr 2012 12:06:55 +0000

I like these concepts and they make great sense. I have been learning lately to be more exact in meaning and not make assumptions. It’s so easy to assume that things will always work a certain way, but what if “they” change something. What if you want to change something?

Being a little more verbose is not a bad thing, and can lead to more flexibility down the road when you least expect it.

Thanks for sharing and giving me a new perspective.

Comment on The Closure Puzzle by Lee Davis

Lee Davis — Mon, 16 Jan 2012 12:52:31 +0000

Despite the explaination of what its doing, am I the only one thinking _why_ is it doing this? Is this really a good feature? What benefits does binding closures to a new scope really give? And do the benefits really outweight the risks?

When managing a highly layered application its comforting to know that solid API’s (which may contain final / private methods) can encapsulate behaviours within themselves. I know the Protected vs Private debate has been iterated more times that i’d like to mention, but I still feel there is a place for private scope and seeing more features that allow you to break it concerns me.

I can appreciate the common retort of “well if someone has access to the code they can do what they want”. This is true, but as release manager you may take comfort in the fact that certain core layers of the application aren’t touched on a commit, and so fully encapsulated behaviours that should never be executed or become visible from outside of that layer.

This along with the reflection API allowing you to break scope fuels the burning question, how relaxed to fundemental OO principles is PHP going to become in the future. Is it even worth using the visibility keywords for > 5.4?

Comment on The Closure Puzzle by Matthew Weier O'Phinney

Matthew Weier O'Phinney — Fri, 13 Jan 2012 18:38:04 +0000

As a succinct explanation: In your example, you’re binding $this to Bat, but the _scope_ is Bar — which means that the closure can only access methods in Bat that are visible to Bar. Since Bar does not extend Bat, that means only public visibility, which means calls to $this->D() will fail.

Comment on Faster Arrays by Lukas

Lukas — Tue, 08 Nov 2011 07:37:03 +0000

Yeah I also do not think that this will be useful in that many cases, but there might be some with an importer. However if we start doing stuff like this “read-only” arrays would likely be something we should look into first.

Comment on Faster Arrays by Michael Stillwell

Michael Stillwell — Mon, 07 Nov 2011 14:38:20 +0000

Have you tried preallocating the regular array with array_fill() or similar? I’d expect “preallocated” to then perform somewhere between “regular” and “fixed.”

Comment on Faster Arrays by Johannes

Johannes — Mon, 07 Nov 2011 13:57:59 +0000

Implementing PDO fetchAll zo return a “fixedArray” would be trivial. I’m not sure whether SPL provides the APIs but even doing it from hand should take less than a day for somebody who knows PHP’s workings.

But a few words of notice: An SplFixedArray is no array but an object. This means it is not using copy semantics but pass-by-handle semantics. It’s not possible to use all array functions etc.

I also wonder how useful this really is. If you care about performance why do you copy the complete result set around? – Read the data and process it dirctly (using some fetch loop etc.) instead of copying it and iterating multiple times (at least one time for the fetchAll and then a second time for processing) Proper optimisation there should bring more benefit than using this not-a-real-array thing.

Comment on Faster Arrays by Davey Shafik

Davey Shafik — Sun, 06 Nov 2011 16:15:49 +0000

It would have to be something like: PDO::FETCH_ALL_MODE_FIXED_ARRAY and you would also need the MODE_ARRAY to undo it. It would also be a PDO option not a fetchAll() option like emulated prepares etc.

Comment on Faster Arrays by David Rogers

David Rogers — Sun, 06 Nov 2011 12:55:51 +0000

Interesting idea for PDOStatement::fetchAll() to return a SplFixedArray for large datasets. Could probably add another class constant to PDO to indicate the desired behavior, i.e. PDO::FETCH_MODE_FIXED_ARRAY.

Comment on New Start by I’m joining EngineYard to work on Orchestra | Ross Duggan

I’m joining EngineYard to work on Orchestra | Ross Duggan — Fri, 07 Oct 2011 10:24:03 +0000

[...] years with Boards.ie, and next week I join Eamon, David, Helgi and Gwoo with Noah, Elizabeth, Davey (and a couple of others) to help kick Orchestra into orbit by joining [...]

Comment on New Start by Davey Shafik

Davey Shafik — Fri, 30 Sep 2011 05:21:14 +0000

They have joined forces with EngineYard as of August 23rd, see more here: http://blog.orchestra.io/post/9291203706/orchestra-engine-yard