[TYPO3-dev] TYPO3 Bug Day coming up this Friday, August 29th

[Christian Kuhn]

Hi,

Franz Holzinger wrote:
> Dmitry Dulepov [typo3] a écrit :
> 
>> Think, would Linux kernel people allow any Linux developer to commit to 
>> the Linux kernel? What will happen then?
> 
> Did you ever see such an error message under LINUX, when everything is 
> blocked?
> 
> LINUX Kernel Fatal Error: Module named "mymodule" was NOT loaded! 
> (linux_extMgm::extPath)
> 
> This never happens under any LINUX, however a normal TYPO3 user will 
> happen to end up here. Nobody would use a LINUX kernel which could cause 
> such a big trouble to users and where not restart could help.

If your refer to the vanilla Linus tree of Linux, there is a well known, 
very monolithic, well structured, long and winding road to get patches 
in. They usually go over Sub or Sub-Sub Maintainers or even through 
separate trees for the testing and fixing process (eg. linux nextgen, 
mm-tree). Even if you have a nice solution to something, but you fucked 
up some tiny bit in this hole process, it might happen that your hole 
happy little project never makes it to vanilla.

Its not as simple as you might think, please refer to [1] for details. 
Linux also changed the hole process several times to get a working 
solution, they still argue about details today. They even hacked up an 
own versioning tool to solve some problems.

And there are still bugs in the kernel. I remember those promise ide 
controllers never where stable for me, the dvb driver still blew up my 
system sometimes and I have a serious regression somewhere in the block 
layer on one of my systems if I use filesystem encryption. Every one of 
those would probably take me years of getting more knowledge to have a 
minor chance to fix them.

Back to TYPO3: IMHO the core is amazingly stable. If you lower the 
barrier to the core any further you will get a totally bloated and 
unstable solution within weeks. I am personally very glad that patches 
need to be reviewed, I still make lots of mistakes and I am very happy 
if a core dev takes a look to stuff or even points errors out.
Serious testing and reviewing is a very time consuming task, you will 
never hear something like "come on lazy devs, get my stuff in" from me! 
I just can do my best to help. Even if there are currently quite some 
pending patches, many of them still have problems and a reason why they 
have not been committed.


Regards
Christian

[1] On submitting kernel patches 
http://ols.fedoraproject.org/OLS/Reprints-2008/kleen-reprint.pdf