[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFS: (RC bugfix, NMU) mindi

Bilal Akhtar wrote:

On 11/28/2010 05:26 PM, Didier 'OdyX' Raboud wrote:
>> FYI, I got this uploaded as NMU some minutes ago:
> Now, you just managed to anger me by doing this.

Hi Bilal, 

let me apologize for the process, as my intent was not to anger you.

I was at a BSP, with several DD's around and we discussed your proposed 
patch; the general agreement was "using conffiles to solve this bug is plain 
wrong". The approach I proposed to you was the result of that "live" 
discussion. I prepared and tested the patch, and got it uploaded, because it 
sounded a saner technical solution to the problem at hand.

> First, what is the harm in adding these files to debian/conffiles?

You should read Policy E.1. It states "The filenames (…), and the files 
referred to should actually exist in the package."

As their name tells, conffiles are supposed to be used for "Automatic 
handling of configuration files by dpkg". As their path tells, the files we 
are talking about in /usr/lib/mindi are not configuration files: if they 
were they would be in /etc. (And actually they should be in /var/lib 
probably, see #605261 and the FHS).

> The only difference from the earlier package would be:
> 1) These files would be removed only on purge.
> 2) There would be no need of a postrm file.

While that might be true (I did not check), I believe (and I was not alone) 
that the technical solution you adopted to solve the problem was just plain 

And as for the "no need of a postrm file", what is the advantage of this ? 
We have maintainer scripts for a reason.

> This was the needed behavior, and this approach was the Debian-policy
> recommended way of getting the job done.

Could you please point me to the Debian Policy chapter that recommends to 
get cache files removed with the conffiles mechanism?

> Secondly, if I had tagged the bug with pending, then I was the person
> who was supposed to fix the bug! And, you just got the thing NMUed
> without any consent from me.

How could I have known without monitoring debian-mentors@ ? You did not send 
the patch to the bugreport…

While I agree that I could have been more patient and that I could have 
taken more time to convince you of the wrongness of the technical solution 
you proposed, I don't think I needed an agreement from you to NMU this: we 
are in freeze, things must get fixed (and I had the opportunity, given that 
I was at a BSP with DD's around).

Furthermore, note that I did not a "zero-day" NMU (immediate upload), but 
got it uploaded to DELAYED/2 : this gives you (trough a DD) the time to 
block the upload from actually reaching the archive.

My NMU will reach the archive in more than one day now [0], so you have one 
day to convince any DD of the superiority of your technical solution over 
the one I got uploaded, get mine rejected and yours uploaded instead.

Finally, note that I asked the Release Managers to remove mindi from 
testing: #605264 .

> Sorry if my mail was too harsh.
> Bilal Akhtar

Don't worry. It was not harsh at all: it just stated your frustration with 
my behaviour, which I can understand. That's completely fine.



[0] http://ftp-master.debian.org/deferred.html

Reply to: