Frans' deletions break to much, make them stop *now*

Koen Kooi koen at dominion.thruhere.net
Thu Aug 19 12:34:27 UTC 2010


Op 19 aug 2010, om 14:23 heeft Holger Hans Peter Freyther het volgende geschreven:

> On 08/19/2010 08:13 PM, Koen Kooi wrote:
> 
>> No, the tinderbox lives inside the firewall, and more importantly, the fallout encompasses binary feeds as well.
> 
> Crap, I was doing some OE stuff the last few days but I have some issue in
> understanding the current problem:
> 	- Is Frans removing stuff actively used by Angstrom-200x or any
>          other distro in OE?

Yes, the most obvious being openssl 0.9.8, which is was the default on in OE. That triggered a build of openssl 1.0.0, which changed soversion. And to make it worse, openssl 0.9.8 and 1.0.0 can't be parallel installed due to both including the same config file. So since a few ipks have been deployed in the feeds all recipes linking against openssl need a PR bump and rebuild. And of course DEPENDS=openssl isn't a good metric :(

> 	- Is Frans removing stuff that your private port is using? Do you
>          think this would change your point?

Right now there is now private port for my department, everything is in .dev, but the tinderbox is internal so we can build crazy stuff before pushing to OE upstream.

> I know that we as TSC have
>          encouraged to remove recipes under certain circumstances. How could
>          we change the guidelines if the above is true? How can people claim
>          library versions?

At least mandate posting a patch for review. And encourage using removal.txt, that's what it's for. I don't mind that stuff gets deleted, I do mind that I can't get any real work done because I'm fixing shit because frans wants to impress his boss with his commit stats.

Have a look at how Martin Jansa appoaches things like this, he post patches, tests things, works with the involved parties, etc. That's a lot different than deleting 500 recipes during the weekend without testing and then claiming the OE quality got better.

regards,

Koen



More information about the tsc mailing list