headers
Silas Silva | 12 Sep 2009 09:42
Picon
Gravatar

sysinst split project

Hello tech-install <at>  and Tim Rightnour,

I was looking for the Google Summer of Code NetBSD projects ideas some
days ago when I saw the sysinst split project here:

http://www.netbsd.org/contrib/soc-projects.html#sysinst_split

Once of the problems I'd had with NetBSD was making its installation
more automatized than it is nowadays, so I thought about working on this
project. I previously had an idea to make a different install tool and,
at work, I already replaced sysinst by a shell script, when we needed an
automatic NetBSD installation for a given equipment.

There is a problem, though, that, since I have a full time job, and I
study at the university at night, I can't submit this project to work on
Google Summer of Code, which requires, IIRC, full time working on the
project. I can't have too much time during the weekdays, but I do have
on the weekend. I can propose a schedule for it.

So, my question is, is it possible for me to work on this and, would I
get reasonable support for you?

Things I have already done:

- Studied sysinst source code;
- Understood the menu generation with menuc;
- Listed important functions (in do_install() function) that are
  candidates to go to the backend process, after likely heavy
  modification;
- Basic specification of a configuration install file for the backend
(Continue reading)

Permalink | Reply | 3 comments |
headers
Eric Haszlakiewicz | 13 Sep 2009 04:32

Re: sysinst split project

On Sat, Sep 12, 2009 at 04:42:34AM -0300, Silas Silva wrote:
> I was looking for the Google Summer of Code NetBSD projects ideas some
> days ago when I saw the sysinst split project here:

> There is a problem, though, that, since I have a full time job, and I
> study at the university at night, I can't submit this project to work on
> Google Summer of Code, which requires, IIRC, full time working on the
> project. I can't have too much time during the weekdays, but I do have
> on the weekend. I can propose a schedule for it.

Well, I don't think the "full time" part is really a hard requirement,
but you would have to be able to get the project done within the SoC
timeline, which may very well require full time work.  It's kind of
irrelevant though, since the 2009 SoC is over (or were you thinking about
this for the 2010 one?).

> So, my question is, is it possible for me to work on this and, would I
> get reasonable support for you?
...snip...

wow, willing to work, organized, and polite too!  I can't speak officially
for the project, but I think most people here would agree with me when I say
"yes, please, work on this!"

With an approach like this I think you'll get plenty of people willing to
help you get this going.  You'll also likely get many opinions, some 
conflicting, about how things should be done, so it may help to have a
little bit of a thick skin. :)  Nevertheless, anyone willing to put out
the effort to work on NetBSD, especially for something like this that
many people have expressed a desire for in the past, will be welcome
(Continue reading)

Permalink | Reply | 2 comments |
headers
Silas Silva | 15 Sep 2009 15:18
Picon
Gravatar

Re: sysinst split project

On Sat, Sep 12, 2009 at 09:32:48PM -0500, Eric Haszlakiewicz wrote:
> With an approach like this I think you'll get plenty of people willing
> to help you get this going.  You'll also likely get many opinions,
> some conflicting, about how things should be done, so it may help to
> have a little bit of a thick skin. :)  Nevertheless, anyone willing to
> put out the effort to work on NetBSD, especially for something like
> this that many people have expressed a desire for in the past, will be
> welcome and encouraged.

Hi Eric! Thank you for your answer. This encourages me a lot.

I'll follow my plan normally, until someone who can speak officially by
the project (not you?) can give me directions, if I'm doing anything
wrong. I'll also propose him a roadmap, before putting this on the
public.

I don't know if this list would be the best place to send questions
related to this project. Maybe personal e-mail and IM networks would be
valuable.

Thanks!

--

-- 
Silas Silva
Permalink | Reply | 1 comment |
headers
Eric Haszlakiewicz | 15 Sep 2009 18:58

Re: sysinst split project

On Tue, Sep 15, 2009 at 10:18:10AM -0300, Silas Silva wrote:
> I don't know if this list would be the best place to send questions
> related to this project. Maybe personal e-mail and IM networks would be
> valuable.

You'll get more of a response if you post to current-users.  There are
a lot more people that read that and not tech-install, even though
they might be interested in things like sysinst.  A quick summary
email (e.g. "I want to work on sysinst, read my proposal on tech-install")
would probably get some people talking.

eric
Permalink | Reply | 0 comments |
headers
Mark Kirby | 15 Sep 2009 20:02
Picon

re: sysinst split project

Hi,

If i remember correctly quite a bit of this project was completed  
during the 2008 gsoc.

The summary page is at http://netbsd-soc.sourceforge.net/projects/install-tool/ 
  and you can view the code from there too.

I have not seen Tim around on the lists for a while now so as eric  
said post to current-users with a pointer to your proposal.

Hope the above link and code is of some help too you.

you may also want to look at the dragonflybsd installer which already  
does all these things and could probably be ported over easier then  
rewriting sysinst. Just a suggestion.

Mark
Permalink | Reply | 501 comments |
headers
Silas Silva | 18 Sep 2009 02:55
Picon
Gravatar

Re: sysinst split project

On Tue, Sep 15, 2009 at 07:02:12PM +0100, Mark Kirby wrote:
> The summary page is at 
> http://netbsd-soc.sourceforge.net/projects/install-tool/ and you can view 
>  the code from there too.

I glanced at it, thanks (going to take a detailed look later). I was
thinking of another syntax for the file, but there are many interested
things there. It looks a good idea to create a lib to parse it.

> Hope the above link and code is of some help too you.

It is and it will be. Thank you very much!

> you may also want to look at the dragonflybsd installer which already  
> does all these things and could probably be ported over easier then  
> rewriting sysinst. Just a suggestion.

Good suggestion. Already downloading it... I'm sure that, if it will not
be ported, a lot of ideas will come from it.

Thank you!

--

-- 
Silas Silva
Permalink | Reply | 501 comments |
headers
David Brownlee | 21 Sep 2009 13:24
Picon

Re: CVS commit: src/distrib

 	(CCing tech-install <at> , as it may be worth continuing there, please
 	drop source-changes-d <at>  from any followup).

 	To anyone tuning in on tech-install <at> , this is regarding merging
 	34 various mini termcap variants under distrib into a single
 	~7K version.

On Mon, 21 Sep 2009, Izumi Tsutsui wrote:

>>> Can't it use src/distrib/common/mktermcap script?
>>
>>  	With the ~same set of terminals (actually dropping 'unknown'),
>>  	mktermcap generates a 7723 byte termcap file, as opposed to the
>>  	6987 byte 'hand crafted' one. Its not really much different,
>>  	but on very space constrained install media maybe thats enough
>>  	to make a difference? I'm not certain which way to go...
>
> IMO, two fragments in ffs (that will be smaller after gzipped
> for install media which often has actuall size restrictions)
> might be worth to reduce maintenance cost in future because
> most these files (and installer itself) are rarely maintained
> as you see.

 	I'm hoping to try to cleanup the unholy mess which is distrib
 	a little. I'd like to provide more common ramdisk generation
 	infrastructure so that most ports can just have a Makefile
 	with half a dozen settings and an 'extras' list file, thus
 	making things a little more maintainable.

--

--
(Continue reading)

Permalink | Reply | 0 comments |

Gmane