[BF] wrong value of cache max-size in workers

Maxim Dounin mdounin at mdounin.ru
Wed Jun 15 16:31:50 UTC 2016


On Wed, Jun 15, 2016 at 03:51:57PM +0200, Sergey Brester wrote:

> >Thanks, looks like a valid win32-related problem. See below for
> >comments about the patch.
> And why you come to this conclusion?

Because the code path in question is only followed on win32.  On 
Unix, workers are forked and just use memory of the master 
process, the shm.exists flag is never set.

> >As this is a win32-only problem, please clearly describe this in
> >commit log. E.g., use "Win32: " prefix in the summary line.
> >Please also describe that this is win32-specific. Alternatively,
> >please use "Cache: " prefix. Also, please use full sentences,
> >including dots.
> Well, I think, by merging you can modify the commit message as you like (I
> renounce the authorship of messsage;)))

It's not going to work, as it's very likely to never happen.

> >More hints can be found here:
> >http://nginx.org/en/docs/contributing_changes.html [1]
> I known this paper (I'm not a first time here) ...
> BTW. With this kind of handling with contributors I understand the people,
> that sometimes or never more want to post anything to the nginx-developers.

The page in question is intended to explain how to prepare patches 
to follow nginx style and make sure patches can be handled with 
minimal waste of time of all parties involved.  Following these 
guidlines maximizes changes of a patch being accepted, and 
minimizes time spent by anyone involved.  Not following basically 
says that you don't care about other's time.

The commit log in question do not follow various guidelines 
explicitly written on the page in question, and that's why I've 
sent you a link.  If you are not following the guildelines 
intentionally - well, that's your choice, and we can do nothing 
with it, but don't expect it will somehow improve handling.

> >This empty line looks unneded for me. YMMV.
> >
> If you'll look at source code you will see this was done exactly as in 2
> another blocks contains this code piece (one above and one bellow). I've
> done it, just thus looks like as in the rest of the code.

Yes, I understand the reasons.  Nevertheless, I would not add it 
there.  If you want to preserve it - that's ok though, as reasons 
are good enough, and that's why I wrote "YMMV".

Maxim Dounin

More information about the nginx-devel mailing list