Discussion:
[9fans] GSoC 2013 - ~24 hours left for org application!
(too old to reply)
a***@9srv.net
2013-03-28 20:38:39 UTC
Permalink
Folks:
We're down to just about 24 hours remaining until org
applicaitons are due for Summer of Code 2013. After tomorrow
evening, the fine folks at Google will grab the application and
some key pages linked from it, such as the ideas page and
student application template, and begin evaluating orgs to
decide who gets to participate.

The application itself is coming along nicely. It's mostly
the same as the past few years, so that's the easy part. The
bigger deal is the linked pages, most particularly the ideas
page. This is a key part of the evaluation process, and is the
thing that gets brought up most in the meetings for rejected
orgs most every year. We've gotten positive feedback on ours
for the past several years, but we still need to have this in really
solid shape.

As of now, our idea page is a bit lighter on the variety of
ideas compared to what we've seen in the past. It would be
really good if folks could think hard about the sorts of projects
they'd be interested in being a mentor for and get those in
there ASAP. It would be especially good to have ideas attached
to a few more prospective mentors.

Remember, putting your name on that page is in no way a
commitment to mentor any particular student/application/project.
We'll take a look at the specific applications we get and evaluate
them at that point. If you're thinking "I'd work on this, but only if
the student looked stellar", then put that in!

The easiest (for me, and given the timing, for us generally)
way to do this is to edit the wiki directly; just follow the template
of the existing examples. If you can't do that, just send me email
with a description of the project (decent detail, please) and an
estimated difficulty, and I'll add it in.

The wiki in general isn't really brought into the application
process, but it will be an important resource for interested
prospective students. If you've been putting off making any
changes there, adding things, fixing out of date info, now would
be a really good time to get a move on that.

Much thanks to all the folks who've submitted ideas so far,
either to me or on the wiki directly. Looking forward to getting
this application done!

Anthony
Steven Stallion
2013-03-28 23:49:51 UTC
Permalink
Damn you Gmail. Damn you.
Steven Stallion
2013-03-28 23:47:31 UTC
Permalink
Hey Anthony,

I'd be happy to sign up as a Mentor again this year. I haven't added this
to the ideas page yet since I've been keeping the project mostly quiet, but
I've been working on porting Plan 9 to the ARM Chromebook. It might not be
the best project for a student since at the moment, it requires a developer
board from south korea that's not the easiest to get a hold of, though cost
is low (around 250). Would it be worth mentioning this on the application?

Cheers,

Steve
Post by a***@9srv.net
We're down to just about 24 hours remaining until org
applicaitons are due for Summer of Code 2013. After tomorrow
evening, the fine folks at Google will grab the application and
some key pages linked from it, such as the ideas page and
student application template, and begin evaluating orgs to
decide who gets to participate.
The application itself is coming along nicely. It's mostly
the same as the past few years, so that's the easy part. The
bigger deal is the linked pages, most particularly the ideas
page. This is a key part of the evaluation process, and is the
thing that gets brought up most in the meetings for rejected
orgs most every year. We've gotten positive feedback on ours
for the past several years, but we still need to have this in really
solid shape.
As of now, our idea page is a bit lighter on the variety of
ideas compared to what we've seen in the past. It would be
really good if folks could think hard about the sorts of projects
they'd be interested in being a mentor for and get those in
there ASAP. It would be especially good to have ideas attached
to a few more prospective mentors.
Remember, putting your name on that page is in no way a
commitment to mentor any particular student/application/project.
We'll take a look at the specific applications we get and evaluate
them at that point. If you're thinking "I'd work on this, but only if
the student looked stellar", then put that in!
The easiest (for me, and given the timing, for us generally)
way to do this is to edit the wiki directly; just follow the template
of the existing examples. If you can't do that, just send me email
with a description of the project (decent detail, please) and an
estimated difficulty, and I'll add it in.
The wiki in general isn't really brought into the application
process, but it will be an important resource for interested
prospective students. If you've been putting off making any
changes there, adding things, fixing out of date info, now would
be a really good time to get a move on that.
Much thanks to all the folks who've submitted ideas so far,
either to me or on the wiki directly. Looking forward to getting
this application done!
Anthony
erik quanstrom
2013-03-28 23:55:24 UTC
Permalink
Post by Steven Stallion
Hey Anthony,
I'd be happy to sign up as a Mentor again this year. I haven't added this
to the ideas page yet since I've been keeping the project mostly quiet, but
I've been working on porting Plan 9 to the ARM Chromebook. It might not be
the best project for a student since at the moment, it requires a developer
board from south korea that's not the easiest to get a hold of, though cost
is low (around 250). Would it be worth mentioning this on the application?
cool!

has the documentation been easy enough to get a hold of?

- erik
Steven Stallion
2013-03-29 00:09:31 UTC
Permalink
'ish. Samsung's documents are somewhat rough, but enough to get a basic
debugging environment setup. I've recently added support to U-Boot for Plan
9 kernel images and submitted a patch (arm-uboot), which adds an option to
the loader to generate a uImage. Eventually I need to write everything up
so others can make use of it. I should have a minimal C runtime this
weekend if all goes well.

So far I believe the only real annoyance is lack of public documentation
for the Mali T604, the synaptics touchpad, and of course dealing with the
Marvell wireless part. These can all be dealt with down the road, hopefully
by more willing individuals!

Cheers,

Steve
Post by Steven Stallion
Post by Steven Stallion
Hey Anthony,
I'd be happy to sign up as a Mentor again this year. I haven't added this
to the ideas page yet since I've been keeping the project mostly quiet,
but
Post by Steven Stallion
I've been working on porting Plan 9 to the ARM Chromebook. It might not
be
Post by Steven Stallion
the best project for a student since at the moment, it requires a
developer
Post by Steven Stallion
board from south korea that's not the easiest to get a hold of, though
cost
Post by Steven Stallion
is low (around 250). Would it be worth mentioning this on the
application?
cool!
has the documentation been easy enough to get a hold of?
- erik
erik quanstrom
2013-03-29 00:15:37 UTC
Permalink
Post by Steven Stallion
So far I believe the only real annoyance is lack of public documentation
for the Mali T604, the synaptics touchpad, and of course dealing with the
Marvell wireless part. These can all be dealt with down the road, hopefully
by more willing individuals!
he! does that leave you land-locked, or have you got
usb ethernet going?

- erik
Steven Stallion
2013-03-29 00:29:27 UTC
Permalink
USB has been working quite well so far. The Arndale board has an onboard
ASIX, which has been working quite well for tftp booting kernel images.
Post by Steven Stallion
Post by Steven Stallion
So far I believe the only real annoyance is lack of public documentation
for the Mali T604, the synaptics touchpad, and of course dealing with the
Marvell wireless part. These can all be dealt with down the road,
hopefully
Post by Steven Stallion
by more willing individuals!
he! does that leave you land-locked, or have you got
usb ethernet going?
- erik
Richard Miller
2013-03-29 09:34:38 UTC
Permalink
Post by Steven Stallion
which adds an option to
the loader to generate a uImage
What's the advantage of using a uImage instead of elf or raw binary?
l***@proxima.alt.za
2013-03-29 09:38:17 UTC
Permalink
Post by Steven Stallion
which adds an option to
the loader to generate a uImage
What's the advantage of using a uImage instead of elf or raw binary?
Popularity? U-boot seems very fond of uImage format, my guess is that
it fits well within their line of expertise. I'm also interested in
the opinion of those in the know.

++L
Charles Forsyth
2013-03-29 10:06:40 UTC
Permalink
Post by l***@proxima.alt.za
Popularity? U-boot seems very fond of uImage format, my guess is that
it fits well within their line of expertise. I'm also interested in
the opinion of those in the know.
Unlike Elf, the format is trivial.
Richard Miller
2013-03-29 10:19:53 UTC
Permalink
Post by Charles Forsyth
Unlike Elf, the format is trivial.
Not as trivial as raw binary, which we currently use for u-boot loading
on most ARMs. The uImage format contains the load and entry address so
you don't have to look up 'man 8 booting' or check the kernel mkfile before
setting up the u-boot script. Are there any other advantages?
Steven Stallion
2013-03-29 16:17:23 UTC
Permalink
U-Boot also performs relocation based on the load address which simplifies
the boot process. I'm working on an XIP kernel, which has cut around 100
LOC out of l.s. It's a small change, but simplifies things considerably.
Post by Richard Miller
Post by Charles Forsyth
Unlike Elf, the format is trivial.
Not as trivial as raw binary, which we currently use for u-boot loading
on most ARMs. The uImage format contains the load and entry address so
you don't have to look up 'man 8 booting' or check the kernel mkfile before
setting up the u-boot script. Are there any other advantages?
Steven Stallion
2013-03-29 16:19:38 UTC
Permalink
I should probably mention, there is no need for much of a U-Boot script
anymore. To boot with a uImage, my bootcmd is quite literally "tftpboot;
bootm". Obviously we still have to do the same dance with plan9.ini, but I
can live with that, though I do have some plans on how to deal with that
later.
Post by Steven Stallion
U-Boot also performs relocation based on the load address which simplifies
the boot process. I'm working on an XIP kernel, which has cut around 100
LOC out of l.s. It's a small change, but simplifies things considerably.
Post by Richard Miller
Post by Charles Forsyth
Unlike Elf, the format is trivial.
Not as trivial as raw binary, which we currently use for u-boot loading
on most ARMs. The uImage format contains the load and entry address so
you don't have to look up 'man 8 booting' or check the kernel mkfile before
setting up the u-boot script. Are there any other advantages?
erik quanstrom
2013-03-30 01:14:59 UTC
Permalink
Post by Steven Stallion
I should probably mention, there is no need for much of a U-Boot script
anymore. To boot with a uImage, my bootcmd is quite literally "tftpboot;
bootm". Obviously we still have to do the same dance with plan9.ini, but I
can live with that, though I do have some plans on how to deal with that
later.
interesting. do you mind sharing which dev kit you are using?
i think the swedish were wrong in dropping "ungoogleable" from
consideration. ☺ ... and not that i have time.

- erik
Steven Stallion
2013-03-30 04:45:36 UTC
Permalink
I've been using an Arndale board. It needed a small bit of rework to
simplify power cycling the board, otherwise it's been quite good.
Post by erik quanstrom
Post by Steven Stallion
I should probably mention, there is no need for much of a U-Boot script
anymore. To boot with a uImage, my bootcmd is quite literally "tftpboot;
bootm". Obviously we still have to do the same dance with plan9.ini, but
I
Post by Steven Stallion
can live with that, though I do have some plans on how to deal with that
later.
interesting. do you mind sharing which dev kit you are using?
i think the swedish were wrong in dropping "ungoogleable" from
consideration. ☺ ... and not that i have time.
- erik
Loading...