[TYPO3-core] RFC: #15814: Feature: Add a concatenate class

[Helmut Hummel]

Hi,

On 25.09.10 22:29, Steffen Kamper wrote:
> 
> i don't want to change compressor for now. Concatenating isn't compress, 
> so this could be done by the concatenate class later.

So the concatenating will be moved from t3lib_Compressor to
t3lib_Concatenator?

>> * you use uniqid() to create a unique filename, which will create a new
>> file on each hit, because the filename is random (file_exists will
>> always fail)
> 
> you have to set the file for your context. I don't expect to concatenate 
> all inclusions, but one for backend, one for rte, one for each extension 
> etc. I don't see a possibility to guess the context, you?

If setting the filename is a requirement, then why don't pass it through
the contructor? The uniqid() is misleading there.

Another possibility to build a hash value over the keys, the inline
contents and the filetime of the files. AFAIK it's done this way in
t3lib_Compressor.

>> * registered files must be relative to PATH_Site, which is problematic
>> if it comes to extensions and the possiblility of different installation
>> locations
> 
> extensions i see as the extRelPath is relative to typo3. What problem do 
> you see with "different installation locations" ?

Ah, OK. I forgot about t3lib_extmgm::extRelPath(). Nevertheless I would
use t3lib_div::getFileAbsFileName() instead of simply prepending
PATH_site to the relative file path.

>> * mimeTypes differ from file extensions. A mimeType is something like
>> "text/plain" not ".txt" So the variable and method names should be changed
> 
> any suggestion? 

$fileExtension?

> I only used a mimeType direct for the fileExtension.

I do not understand this sentence.

> thx. As i wrote i first want to collect exact these comments to create a 
> new patch.

Wouldn't be the dev list a better place for proposals in such an early
stage?

Regards Helmut