Re: [gentoo-dev] Please use land-misc herd where appropriate! No no-herd madness!!!

[George Shapovalov]

To ease off foser's comments I'll throw a few more words :).

First I'd like to state that *personally* (emphasis to avoid accusations that 
I am forcing non-existant policy ;)) I am against "registering" no-herd in 
herds.xml. This does not bring us any close to the accountability which is a 
primary goal of all this metadata introduction, but rather makes this 
temporary "solution" more permanent (not like its too transient atm :)).
However unfortunately not every situation can be easily resolved, but I'd 
rather have us come up with some better solution..

Ok, condensing your message I see you mentioned two situations where you feel 
no-herd is necessary:
> when an ebuild does not fit into any existing herds or when a
> maintainer is not apart of what would perhaps be a fitting herd.

While 1st situation is truly unaccounted for and gives the most trouble, the 
2nd one is covered and was specifically discussed (back all that time). In 
this case (package belongs to some herd but has a maintainer not willing to 
be in that herd) metadata.xml is supposed to list this herd and have a 
separate maintainer entry for this dev who, in turn, would not be listed in 
that particular herd in herds.xml. Bug wranglers are supposed then to assign 
all package-related bugs to this maintainer and (may be optionally, I am not 
totally sure on this one) CC the herd alias.

As foser pointed out, if the package is not yet added to this herd this should 
first be discussed with that herd maintainers. Since we are talking about a 
situation when this package conceptually falls into the corresponding 
category, herd devs shouldn't be opposed with the understanding that the 
proposing dev will serve as its maintainer (and bug-wrangling happens as 
described).

Actually, the 1st situation might be attacked by introducing some more generic 
herd (e.g. x11-misc, just follow the categories) and individual devs list 
themselves in herds.xml with a <role> clause or just have a single fall-back 
person and add themselves to individual metadata.xml's as maintainers. The 
utility of this approach is limited to some basic accountability and I 
realise it may be hard to sign-up people to such groups. But that's at least 
an attempt :). May be somebody would have a better idea on how to cover such 
situation (not involving a non-existant herd and thus dropping any hope to 
get a fall-back coverage)?

George


--
gentoo-dev@g.o mailing list