This is the type of response I'm very pleased to elicit. Both
technical issues and political ones need to be in the open to make
sure we don't drive down a dead-end road, so I'd like to encourage
matching skills to the various tasks, so we have some idea of what
progress is possible.
I know that Anthony has added massive value to the project and I doubt
anyone else will catch up for a while, I'd like to know where he is at
and whether he'd like some help taking things further. I am under the
impression that he's had other pressures and perhaps he'd like others
to volunteer to relieve him. I know I was reluctant to take the 9vx
stuff off his shoulders, but now that much time has passed and I am
not as frantic, maybe I can help there.
This is not intended to be a comprehensive list, there is much going
on that I am not familiar with or even aware of and I hate making a
fool of myself. Gorka and I were doing moderately well with the ARM
port, but I was distracted and I'm sure Gorka had his own issues to
attend to. I'd like to go back there, but it will take me a few weeks
to get back in some sort of sync, so I'd rather leave this for a while
and focus on the 386 port for now, chase away the threat of having the
distribution dropped.
At the same time, I would like to encourage forsyth to push ahead with
the fix for 5c that allows switch labels to be vlongs, as that is a
real thorn. If somebody can help him, we'd all benefit.
I would also very much like a list of Plan 9-specific packages, I know
that good work has been done in that direction, but it's hard to
locate: this could easily cause duplication, something that detracts
from the availability of resources and may give rise to conflict.
I can't think of anything else, but I'm sure there is plenty more that
can be contributed. My thinking at one stage was to use the Go Issue
Tracker and I know that the Go Team would be OK with that as long as
we label the issues clearly as Plan 9 specific. But maybe we need an
issue tracker of our own?
Lucio.
On 11/15/13, Nick Owens <
misc...@offblast.org> wrote:
> first off, this is awesome. i figured something like this might happen
> once 1.2 got pushed out the door.
>
> second, i would happily offer a builder machine, although all my systems
> run 9front and the servers i have access to run KVM. i have access to a
> ganeti installation at iocoop in the south bay area that could be used,
> and if necessary i can install bell labs' distribution.
>
> some trouble points that come to mind -
>
> - signal handling (especially writing to closed pipes, which is
> unhandled)
> - os api is very poor, checks certain filesystems errors by string
> comparisons
> - bootstrap uses system libbio, which requires patching to have the
> same functionality as go's libbio
> - sse instructions still seem to be a problem, i just disable calling
> cpuid alltogether.
>
> i can elaborate further on the ml.
>
> nick
>
>