Re: CVS commit: pkgsrc/net/nextcloud-client

[Greg Troxel <gdt%lexort.com@localhost>]

Hauke Fath <hf%spg.tu-darmstadt.de@localhost> writes:

> On Mon, 17 Nov 2025 09:08:33 -0500, Greg Troxel wrote:
>> "Adam Ciarcinski" <adam%netbsd.org@localhost> writes:
>> 
>>> Log Message:
>>> nextcloud-client: updated to 4.0.1
>> 
>> Hauke just [...] noted in the commit message of update to 3.17.3:
>> 
>>   [...]
>>   While here, update to 3.17.3 - 4.x is a bit of a mess, still.
>> 
>> If there is valid concern that 4.x is not baked, we shouldn't impose it
>> on users.  Absent solid evidence that it's ok, I think we should be on
>> 3.17.3.
>
> To elaborate on my comment: Like many other German universities, TU 
> Darmstadt has recently switched their cloud storage offering to a 
> centrally managed Nextcloud server. As such I have a bit of client-side 
> support on my plate.
>
> The v4 client was rolled out without further announcement or prompt to 
> anybody who has automatic updates activated. 4.0.0 went and created 
> recursive folder structures in your sandbox that 4.0.1 does nothing 
> about, and that need to be cleaned up manually. I rolled back to 
> v3.17.3 all client installations that I had to touch, and disabled 
> automatic updates. 
>
> Upstream's handling of bugs does not appear to be substantially faster 
> than Mozilla's - the tickets fill up with user complaints, while the 
> developers seem to be busy fishin'.
>
> Given the imminent branch, I decided to stay on the safe 3.x side, 
> instead of adding to (potential) last-minute breakage. 

That sounds like a serious bug, and I think an immediate revert is
warranted.

One of the fundamentally broken things about pkgsrc is that when 9
people think an update shouldn't happen, and 1 does, it happens.  I am
not advocating for a process change this minute in terms of review of
changes, but I think that it's pretty obvious we should treat
post-commit objections as equivalent to objetions during the pre-merge
review that we didn't have.

Please add a comment that updating to 4 is not ok.  In general, while I
think pkgsrc developers should be cautious about big updates, I think
it's also good that when someone decides not to update, that this
decision and rationale be included.