[U-Boot] [ANN] U-Boot v2018.11 delayed a day

[Boris Brezillon]

Hello Wolfgang,

On Tue, 13 Nov 2018 21:36:17 +0530
Jagan Teki <jagan at amarulasolutions.com> wrote:

> On Tue, Nov 13, 2018 at 9:27 PM Wolfgang Denk <wd at denx.de> wrote:
> >
> > Dear Jagan,
> >
> > In message <CAMty3ZC_u=cigEYqj+fsurU8x=pqLCQ8HbVEsrA1qVFvR3ciWQ at mail.gmail.com> you wrote:  
> > >  
> > > > That patch is buggy and needs to be respun and retested. I don't think
> > > > you can push known buggy patch into a release as a fix.  
> > >
> > > True, ie what my yesterday plan [1] but this seems to break some know
> > > functionalities in MTD, can't we go with some immediate fix for this?  
> >
> > Your request makes no sense.  You just saw what "immediate fixes"
> > mean: adding untested code, which will make things not better, but
> > rahter worse, as you _think_ you have fixed something - while in
> > fact you have done the opposite.
> >
> > It is _always_ a Good Thing to allow for sufficient time for
> > testing.  Even if this delays a release.  
> 
> These changes were fully tested, not an immediate changes. What I mean
> a immediate fix here is not for the series, for the comment mentioned
> by Marek about patch 1/5.

More details about the current situation.

First of all, if you want to blame someone, it should be me not Jagan,
since I'm the one who insistently asked Jagan to apply the patch series
and send a PR to Tom. Now some explanation on what this series fixes.

The bug it fixes has been reported by Stefan Roese here. It was fixed
here [2]. It's also been tested by Stefan, I then fixed build failures
reported by Jagan [3] and finally the problem reported Marek yesterday
[4].

So yes, it's unfortunate that is took us two weeks to find those
issues, but, to be honest, I don't think we would have gotten more
reviews by delaying the series to the next release. Instead, the
request to merge this patchset in v2018.11 did trigger a review from
Marek who found a problem in the code.

Regarding the lack of testing, I understand your concern, but look at
the amount of config options we have for MTD or the various way we can
retrieve the mtdparts/mtdids info. I don't think we can easily test all
of these, or at least not in a reasonable amount of time. I'd love to
be proven wrong though.

Now, if [4] is a problem and you think it shouldn't be merged, let's not
delay the release for that. UBI will be buggy on setups with a
spi-nand, but I think we can live with that since it's a new feature
anyway.

Regards,

Boris

[1]http://patchwork.ozlabs.org/patch/990696/
[2]http://patchwork.ozlabs.org/project/uboot/list/?series=73381&archive=both&state=*
[3]http://patchwork.ozlabs.org/project/uboot/list/?series=75389&archive=both&state=*
[4]http://patchwork.ozlabs.org/project/uboot/list/?series=75687