> From: Marcus Börger [mailto:marcus.boerger@t-online.de]
> Sent: Thursday, September 04, 2003 5:13 PM
> referring to paragraph 6 of our CODING_STYLES i changed my own spl
> extension
> and parts of SQLite which interact with spl.
>
> [6] Method names follow the 'studlyCaps' (also referred to as 'bumpy
case'
> or 'camel caps') naming convention, with care taken to minimize
the
> letter count. The initial letter of the name is lowercase, and
each
> letter that starts a new 'word' is capitalized.
>
> Good:
> 'connect()'
> 'getData()'
> 'buildSomeWidget()'
>
> Bad:
> 'get_Data()'
> 'buildsomewidget'
> 'getI()'
>
> This change immediatly caused stormy discussion on IRC.
> Here is what i think:
> * nobodycanreadysudlycapsmethods <- can you read that? That is what
our
> error messages and the reflection output will contain instead of the
> intended nobodyCanReadySudlyCapsMethods what isn't much better.
> Anyway PHP isn't simply capable of handling sudlyCaps since we need
to
> input method names in lowercase.
>
> * When moving extensions from PHP4 to PHP5 we would sooner or later
add
> newer features to their classes just like i did with SQLite. This
does
> result in a mixture of the old and new naming conversion which is
> annoying.
>
> As the only consequence i think we should remove that part 6 from our
CS.
PEAR has had their error messages all lowered cased since it's creation.
I can only speak for the last 1,5 years were I dont remember this ever
being an issue that was brought up. Maybe something to consider is the
fact that in OO you dont have to have the "ext_" prefix and therefore
most method names should only consist of 2 in rare cases 3 words.
I just want to clarify what this means for PEAR.
It means we have to change our CS, because otherwise we either have a
mixed CS which sort of defeats the purpose or we have to wrap every
object that we want to extend that is provided by an extension.
This obviously means a mess beyond believe for PEAR.
I know that there is no way to have studlyCaps being preserved in error
messages without a huge performance hit since we have case insensitivity
for function/method names. However aside from the other reasons I stated
when I initially proposed this, I just wanted to make clear what this
means.
So is this really such a huge issue?
Regards,
Lukas