[SM-Discuss] Project Lead Speech
dkowis at shlrm.org
Sat Mar 5 18:18:13 EST 2011
tl;dr: I HAS IDEARS AND WANT TO CHANGE SOME THINGS. VOET FOR ME! KTHX
-- But seriously, read it.
<Personalized Subject Time Here> Mage,
SourceMage is my favorite distro. Unfortunately, it is also the distro
that causes much rage for me. I have been unable to use SMGL as my
desktop for quite a while now. I'd boot up my linux, want to do some
work, "Oh, need to update <spell> to do whatever." Do the update. 1-2
hours later, my system is either broken, or Firefox no longer works. Or
maybe X has crapt it's pants.
For servers, there is no equal to Source Mage for customization. Updates
are far less disastrous in that arena. I run SMGL as my Xen domain0, and
will continue to do so for the foreseeable future. I have picked out
some Fedora VMs to run some things. I should probably use CentOS or
something that doesn't update often, once they get the items I want in
the distro. Not because I dislike SourceMage, but because I kept running
into problems using things. Castfs would occasionally barf on me and eat
up 100% of the CPU time. Often an update to glibc, or gcc would
irrevocably break my system. Updating server software, like Apache,
Proftpd, Postfix, Postgresql, or MySQL (Yeah, I should switch to mariadb
[fsck oracle]) all work wonderfully. Rarely have I run into an issue
when updating those spells.
Now that I've laid out my gripes and praises for SMGL you might be
wondering why I even accepted the Project Lead nomination if I've got so
many things that I don't like about SMGL and when I use it less. I
/want/ to use SMGL. I miss it. But I can "just use" Fedora easier. I
want to rectify this situation. I want to be able to "just use" SMGL
more. I want to improve the spell writing process so that it's easier to
handle simple version bumps. I want to collect more feedback from the
users automatically, detailing what architecture they're running, and
what spells they're installing, and whether they succeeded or failed.
This will give us, the crafters of the distro, better insight into
what's going on. Simplifying the spell writing process, I believe, will
help us spend less time on version updates. Ultimate goal would be
something like having a bot that listens to the sm-spell-updates list
and can automatically update a spell's version, run a test cast, and
commit the spell into a branch or a patch, so that it's subject to
developer approval. We'd have a software system where the test cast
would include installed files, and whatever other data we wanted
collected for us automatically, so that we don't have to do all that
work manually, and our precious developer time can be better spent on
Secondly, and possibly more importantly, I want to lower the barrier to
entry for developing on Sorcery. The last announcement for a stable
sorcery release was almost two years ago. I couldn't find an
announcement for 1.14.4. I'd like to develop a roadmap of features that
we want to have. I want to have them defined using some kind of testing
framework, so we know when these features are accomplished, and so we
can easily verify that we've not made any regressions. Bringing down the
barrier to entry for development on our core system is important. We
need to allow developers to provide enhancements and bug fixes whilst
still ensuring that nothing is being broken, and we need to be able to
do it without all the verification resting on one person.
Thirdly, I want to clean up our bug tracker and our presence on the
World Wide Web(tm). If this involves a cutover to a completely different
bug tracker, or just setting up a second bugzilla and exporting the open
bugs we actually care about. I would like to have better integration
with the bug tracker, the source repository, the wiki, and static web
pages. Something like Redmine or Trac might be nice. We could probably
achieve the same thing with individual pieces of software, but given how
much we use the website, I think it'd be quicker and easier just to go
for an integrated solution and then solve individual issues if/when we
come to them. At the least, I want to pick out a different wiki (I know,
*again*) that has a suitable spam solution, or integrates better with
some kind of centralized account management.
I want to get back into the habit of holding developer meetings again.
We'll post our agenda on the wiki, and everyone can edit the agenda so
anyone can bring up anything, as long as it's planned out on the agenda.
This will help us keep our meeting short, sweet, and people will know
what is going to go on and whether or not they need to be there.
Obviously, I'll try to be at most all the meetings, or have an Assistant
Project Lead designated to act in my stead. I think meetings once a
quarter would be a good goal to start with.
I'd like to have some kind of incentive system set up for people that
volunteer to do the nasty work that no one wants to do, or isn't any
fun. I'm uncertain how this would come into being. Recommendations from
an Open Source community might carry weight in some professional
circles. Or perhaps Developer of the Week status on the website. I'm
uncertain how this would evolve out, but I want to help motivate people
to do some of the things that are hard, like say, replacing
simpleinit-msb, or unifying the gcc spells again along with rebuilds and
such. The real tough projects that are more than "I bumped the version
and added a configure flag." A bit of publicity you can add to your
LinkedIn account, Facebook, or <Social networking Thinger Here>.
I think that's probably a pretty ambitious set of goals:
* Configuration Feedback System
* Improved Spell Config
* Automated Testing
* Lower Barrier for Sorcery Development
* Improved Web Presence
* Developer Meetings
* Incentive System
Obviously, I can't do all these myself, I will need your willingness and
cooperation to do things you may not find fun, and maybe even some
things you don't necessarily agree with. I'm easy on deadlines, but I
want well defined goals. We can always push something off until later.
As for well defined goals: I don't like things like "it works on my
machine." I want some kind of test that can be run by everyone, manually
as a last resort. I want to build confidence in our distro, and I'll
need your help to do it.
Of course, if you think Source Mage is the best thing ever (it *is*),
but you don't want *anything* to change, a vote in my favor might not be
the best option. I want to keep the core of Source Mage the same:
Systems Admin choice rules supreme.
I want to make it easier for us to give the SysAdmins the best spells
with the most options, and the safest upgrade path possible. No small task.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 728 bytes
Desc: OpenPGP digital signature
Url : http://lists.ibiblio.org/pipermail/sm-discuss/attachments/20110305/0f4c493a/attachment.bin
More information about the SM-Discuss