[TYPO3-dev] Proper DIFFs vs. Less work for Devs

[Ernesto Baschny [cron IT]]

Marcus Krause wrote: on 18.05.2009 14:06:
> Mathias Schreiber [wmdb >] schrieb am 05/18/2009 01:13 PM Uhr:
>> Ernesto Baschny [cron IT] schrieb:
>>> Mathias Schreiber [wmdb >] wrote: on 18.05.2009 12:21:
>>>
>>>>> I know it's pear and all but I thought I'd send the link anyway and you
>>>>> can have a look if you can use something from it:
>>>>>
>>>>> http://pear.php.net/package/Text_Diff/docs/latest/Text_Diff/Text_Diff.html
>>>>>
>>>> There are quite a lot of diff tools out, but i need the other way
>>>> around.
>>>> Taking our experiences from T3DD into acount I need to stick as close as
>>>> possible to pure PHP/TYPO3 API as I can.
>>>>
>>>> So I will basically port the patch command from C to PHP (hoorray) so I
>>>> get the same behaviour.
>>>>
>>>> But of course I am open for other approaches, thanks for the link ;-)
>>> I am not sure what you are intending. You should always do a "svn
>>> update" before you create a patch. Than it doesn't matter what tool you
>>> use. Why would a PHP-based tool work better than "svn diff"?
>> I am talking about "patch" not "diff".
>> What I try to do is simple:
>> Build a BE Module
>> Query Mantis via SOAP
>> Find bugs pending in core list
>> Click the bug, get the teams-core thread
>> Select the patch
>> Patch your system
>> Do the functional check
>> Reply into the mailing list
> 
> Great concept.
> 
> My suggestion:
> Why would you want to code the most complex stuff from the beginning?
> 
> Do the actual patching with a TYPO3 service! Start with a service that
> is basically a wrapper around native patch binary. I usually have patch
> binary installed and would be willing to make it available for TYPO3.
> 
> If you or we all receive only few feedback, then this might indicate
> that patch binary is not available for everybody who is willing to use
> your extension. Only then I would add a second service that simulates
> patch behavior in PHP.


Mattes, now I've got your point! Great idea too, sounds really useful!

But I also would go with Marcus' idea of first requiring an installed
"patch" binary and coding the other "funky" stuff around that.

Cheers,
Ernesto