kicad-developers team mailing list archive

[Seth Hillbrand <seth@xxxxxxxxxxxxx>]

Am Do., 4. Okt. 2018 um 10:20 Uhr schrieb Wayne Stambaugh <
stambaughw@xxxxxxxxx>:

> On 10/4/2018 1:11 PM, Seth Hillbrand wrote:
> > Am Do., 4. Okt. 2018 um 09:58 Uhr schrieb Wayne Stambaugh
> > <stambaughw@xxxxxxxxx <mailto:stambaughw@xxxxxxxxx>>:
> >
> >     Still waiting for the macos fix to tag 5.0.1.
> >
> >
> > The 5.0.1 bug doesn't affect 5.1 as far as I can tell.
>
> It's the delta I'm worried about.  This is a significant change of
> non-UI code.  If we would have a critical eeschema bug in the 5.0 branch
> between now and 5.0.1 release, it makes the probability of merge
> conflicts just that much greater.  If the macos fix is going to drag on
> too long, them maybe we could make the argument to merge it.  I figured
> we would already have tagged 5.0.1 by now.  Are we having fun yet? ;)
>

While true, the potential conflict will exist in one direction or the
other.  Either hard to merge the fix into 5.1 or harder to rebase the GAL
branch.  The Mac bug is an unfortunately slow process as it only shows when
building using 10.11, which we only have in a vm.  Because the bug only
shows up when building on the vm and then testing in Mojave, each test
takes ~1 hour of compile time.  So. Much. Fun. :)  We're now looking at
testing the alternate approach but this is about 8 hours of regression
testing spread over all MacOS versions.  I have about 1-2 hours per night,
max.  Which is to say, there's about 1 week of testing left before the Mac
version is ready.  And that assumes we find no regressions.

Going forward, I think we might look at an official policy of only
supporting OSs that are actively supported by their originators.  That
would week out some older systems but reduce the testing load.  We could
still provide compatible binaries but with no bug fixes if they are
system-specific.

-S