I'm disappointed with TC's decision. IMHO it is lacking depth of judgement.
The original problem I reported is absolutism - "vendor everything".
And TC apparently considered only that without trying to find the balance.
What are the reasons for vendoring stable well maintained library like
Logrus? None. That system library can by used easily with hardly any burden
to maintenance. Same can be said probably about ~100 other libraries.
We could do so much better with recommending at least to un-vendor trivial
cases of 3rd party code...
> The Committee declines to overrule the maintainer and accepts the level
> of vendoring used in the 'kubernetes' source package. We also decline
> to intervene in bugs requesting that vendored components in the
> 'kubernetes' source package be installed in binary packages in such a
> way that other packages can make use of them.
> Our consensus is that Kubernetes ought to be considered special in the
> same way that Firefox is considered special -- we treat the package
> differently from most other source packages because (i) it is very large
> and complex, and (ii) upstream has significantly more resources to keep
> all those moving parts up-to-date than Debian does.
That is based on misleading Elana's comparison that I've answered
(regretfully too late) in the other email...
> Our most basic reason for this point of view is that given how much
> fewer resources we have than upstream Kubernetes, it is not feasible for
> Kubernetes to make it into a stable release of Debian unless we take an
> approach like this one.
IMHO Kubernetes is not suitable for "stable" either way.
> The same goes for the possibility of providing
> security support. And given that a strategy for security support is
> available, we do not see any reason why having the Kubernetes bundle in
> a stable release alongside other copies of its vendored dependencies is
> worse than not having Kubernetes in a stable release at all.
It is not suppose to be "all or nothing" approach. We are talking about 3rd
party code, not maintained by Kubernetes. At least some of it is perfectly
feasible to take from packaged libraries instead of messy upstream bundle.
> It should be noted that we think the greater resources of upstream
Let's not praise greater resources unnecessarily. Upstream don't care
about good versioning practices and even automatically closes bugs for
inactivity without addressing them (e.g. https://github.com/kubernetes/
kubernetes/issues/17077).
> is
> relevant not only to keeping on top of patches and security fixes, but
> also to license compliance. It is our belief that there is no reason at
> the present time to be concerned that non-DFSG material would find its
> way into the package, because Kubernetes upstream care about ensuring
> that all vendored dependencies are free software, and they have the
> resources to check.
Since when we are delegating DFSG compliance to upstream based on faith??
FYI in case of "cadvisor" - a required Kubernetes component, DFSG compliance
was difficult to achieve due to pre-minified source-less 3rd party javascript
files...
It was a hell of an effort to get Kubernetes through NEW process. Apparently
I should not have bothered with all that because upstream is so "good" (it
wasn't).
Did you ask ftp-masters opinion/position on that?? Are they OK with volume
of needless garbage in vendored bundle? Would they allow that before TC's
conclusion?
> This could change in the future and it may not be
> true for other upstreams.
I've spent about a year of effort for introducing Kubernetes to Debian - yes,
I had that much faith in Kubernetes but ultimately the effort lead to
disappointment in the technology and loss of confidence in upstream's
ability to maintain the project. But in the course of the effort many new
libraries were introduced as packages, the Debian Golang ecosystem was
stabilised and we've gained a valuable experience with maintaining a complex
Golang software. The hard work was accomplished and it was only the matter or
maintaining Kubernetes properly...
The TC's decision tells me that the effort was largely wasted because it was
"too hard" to do things this way... Naturally the sloppy way is so much
easier...
:(
--
Best wishes,
Dmitry Smirnov
GPG key : 4096R/52B6BBD953968D1B
---
There are occasions when it pays better to fight and be beaten than not to
fight at all.
-- George Orwell
---
"Increased Risk of Noninfluenza Respiratory Virus Infections Associated
With Receipt of Inactivated Influenza Vaccine".
-- https://pubmed.ncbi.nlm.nih.gov/22423139/
Attachment:
signature.asc
Description: This is a digitally signed message part.