[Fedora-packaging] Proposal to reduce anti-bundling requirements

[Jóhann B. Guðmundsson]

On 09/11/2015 07:59 PM, Adam Williamson wrote:
> On Fri, 2015-09-11 at 19:32 +0000, Jóhann B. Guðmundsson wrote:
>> On 09/11/2015 07:25 PM, Adam Williamson wrote:
>>> In a world where bundling was allowed, the package would likely
>>> have
>>> been approved on initial review; the only significant issues found
>>> in
>>> review were bundling-related. There are a couple of trivial issues
>>> noted in #c7, but those would have been literally 10-second fixes.
>> Would have, could have, should have. . .
>>
>> So let's play that game ;)
>>
>> If all the related review request had been completed in timely
>> fashion
>> he would have never given up on un-bundling it.
>>
>> I'm not saying you are wrong but I'm not saying that you are
>> entirely
>> correct in your assumption either what I'm saying is that there are
>> multiple factors at play here.
> OK, so let's talk about review requests! Clearly, we have more review
> requests than we can handle, hence there's a giant backlog, hence
> general sadness.

Indeed

>
> How many of those review requests, do you think, are for tiny bundled
> libraries that will probably only ever be used by at most two packages
> (probably only one)? Wouldn't we have much less of a backlog if we
> didn't have to do all those unbundling requests? ;)

I would argue not since the backlog increased into becoming unmanageable 
at the time when the decision was made  to make it easer to submit and 
include components in Fedora ( which was around FC6  )

>
> Again, I don't actually think the answer here is "screw it, let's
> bundle everything" - but I do believe it's reasonable to say that the
> strict no-bundling policy is causing a lot of fairly pointless work

I hardly call people dedicating their free time into something they 
believe in is the right course of action as pointless work but OK.

> (I'm really not sure unbundling tiny crappy PHP 'libraries' that have
> no sane upstream maintenance policy in any case has ever actually
> benefitted anyone, anywhere, very much), and there are definitely
> cases where it is a primary cause of people abandoning review requests
> or simply not bothering to submit them because the required unbundling
> would be way too much work to be worthwhile. There is a *genuine non-
> zero cost* to the unbundling policy, which is all the OP was
> suggesting.

There exist cases on both sides but that argument is irrelevant in the 
end since all roads lead to the bigger question which has already been 
answered by the board.
( people will first debate where to draw the line if that discussion 
wont be killed in birth but in the end they end up with the same 
question as has already been answered )

JBG