The Gentoo developers are doing one of their regular-ish status report runs. I’ve summarised the reports at http://en.gentoo-wiki.com/wiki/Projects_Status
This is a great way to get an idea of what different development teams are currently working on and where they need help.
Direct links to the original thread and project homepages can be found on the above mentioned article. There’s a forum thread at http://forums.gentoo.org/viewtopic-t-763205.html
Just had an interesting conversation with #gentoo-dev on supporting upgrades from old installs. There was also a very similar discussion on the gentoo-dev mailing list recently on Deprecating EAPI 0. Recently there have been a number of users on the forums upgrading from installs that only support EAPI 0. To achieve this currently, you have to manually intervene. There are various discussions on exact commands to run on the Gentoo Forums, so I won’t go into them again here.
For those that don’t know, EAPI is basically a specification of features available in the package manager. This allows Gentoo to support multiple package managers (Portage, Paludis, Pkgcore, etc) and still move the packaging format forward in a sensible manner.
It was initially suggested that a given set of packages (the trinary of Portage, Bash and Python seems to be the key, in my experience) and their required dependencies should be held back in terms of EAPI required to install.
At the end of the day, I think Gentoo needs to decide how old a system can get before an upgrade can no longer be officially supported. Currently I reckon you need to upgrade at least @system once a year to be able to upgrade without running into too many major issues. I also suspect that the maximum period that many users, particularly those in business environments, would like to have is longer than this (18 to 24 months).
As a method to determine how old a system can get before it is no longer sanely upgradeable, it was proposed that installs be set up, starting at about 1 year old (2008.0 could be used for this as it’s now almost a year old) and jumping back in set periods (perhaps 3 to 6 months). You then try to upgrade these installs to current stable, noting down the problems you encounter and how to resolve them.
Using the notes generated from the above tests, you can then, I think, set a benchmark for how old a system can be before it can no longer be sanely upgraded. In addition, it would be great if some (possibly (semi) official) documentation could be provided on the best manner to upgrade from each of the tested points (it may well be possible to generalise this documentation).
Once you have this information, you can then determine how old an EAPI needs to be supported and what intermediate versions of essential packages may need to provided. The exact method for then handling old systems can then be worked out. There are several possibilities, which include:
a) Hold back certain packages in terms of EAPI
b) Provide multiple revisions of certain packages at different EAPI levels (exactly how you’d version these was not discussed, but I suspect you could have EAPI 0 (or whatever the oldest “supported” EAPI is) as -r0 and EAPI 2 (or whatever the current EAPI is) as -r1.
Edit: I’ve also posted a poll on the forums to help gauge how long some users need the “supported upgrade” period to be.
The KDE project has unveiled their Brainstorm project. The basic concept being that users submit suggestions for improvements to KDE. Jeff Spaleta has some interesting comments on the project.
At a glance, Brainstorm looks to be an interesting project and I think that as well as focusing the time of existing developers it could help to encourage others to get involved by giving them specific features they can implement and submit.
Certainly something to keep an eye on, I think.
Fedora Weekly Comic has an interesting take on the idea.
By way of Jeff’s post, I came across the Kernel oops bugcount site, which the Kernel devs are using to focus their efforts.
I tried and failed to get the new design working with my XSLT setup, so I gave up and put all the content under Wordpress. To save myself time I spring cleaned a lot of the old stuff. There may be a couple more bits I want to add back from the old site, but this is mostly it now.
All links should be maintained thanks to some apache redirect wizardry. If you do find anything broken, let me know.
In related news SVN sucks: “cd blog; svn mv blog ../” (everything’s fine) “svn ci” (frack off – file already exists in repo).
Rinse, repeat. Works fine. Why? Just: Why? -.-
I’ve just performed a long overdue blog upgrade. It looks like my current theme doesn’t support all the features, and there’s no newer version that I can find. Looks like I’ll be putting a new theme on my site at some point soonish, tho it doesn’t look urgent as everything appears to work as intended. If you find any issues, let me know =)
Edit: I’ve now also switched to a new theme. The theme will be going sitewide shortly. There’s a couple of things I might change about the theme, but one of my criteria was minimal changes to get it right. The only major changes I made were to make it stretch to fit the window and not use fixed size fonts. Other than that there’s just some changes to the sidebar – added my sites menu and the feeds links. Removed the recent posts / comments.
At some point I may decide to switch from categories to tags too.
As some people have already heard, the Gentoo Perl developer team is currently desperately in need of new members to help maintain existing and test new versions of Perl and its modules and libraries. If you’re an existing user, don’t worry – the current packages are still being maintained and security updates are being done.
If you’re interested in Perl and would like to help maintain and develop the Perl packages in Gentoo, please contact the Gentoo Perl mailing list or shout on #gentoo-perl on Freenode.
To get into development you don’t need any prior knowledge or epxerience, tho in this case obviously experience developing with Perl will help. There’s no lower limit to your commitment time-wise, so you can help even if you only have a few hours a month to spare.
Specific tasks currently being undertaken by the Perl team include testing of Perl 5.10 and developing packages for the Catalyst framework, as well as general package updates.
If you’re interested in other areas of Gentoo Development, try contacting the specific projects or just shout in #gentoo-bugs on Freenode.
Hello to Planet Larry readers. o/
It seems that I’m not the only one that thinks too much Gentoo development work gets blackholed (as one developer puts it) into IRC, as a recent discussion on the gentoo-devel list was started by a developer. To summarise it looks like there might be quite a few devs who think the same.
One of the thoughts I had on this was to why more developers don’t update project pages. Having done some work with the Gentoo documentation / website, the thought occurs that many do not want to deal with GuideXML, which all the Gentoo web pages are written in. I aired this on #gentoo-chat and bonsaikitten mentioned that most developers probably don’t think to ask the documentation team to convert stuff for them. This is certainly a solution. I wonder if it’s only a stop-gap tho. Something to investigate would be whether developers would prefer project pages to be static HTML or even CMS based. While all this XML-ness can be nice for things like the handbook which do want to be produced easily in multiple formats, I personally think it’s OTT to use it across the whole website.
As with my previous blog post on a new format for the newsletter, the initial cost of setting up a CMS or perhaps even something like Redmine so that devs can easily update project pages would be high, but as I said in the -devel list thread, I believe increased information from the devs would result in higher contributions and possibly even more developers, so it would be well worth it.
In other news, a bit of investigation shows that the GMN is not completely dead. The devs are currently working hard on it, but don’t necessarily expect any monthly newsletters in your inbox too soon. I shall say no more or nightmorph will be sending out his assassins. =D
I don’t usually go in for these meme-type things, but this one’s quite good.
Go to Google and type in your first name and the phrase. Copy and paste the first full sentence/phrase you get that makes sense.
Q: Type in “[your name] needs” in the Google search.
A: Allen needs to know first and foremost WHO African-American people are.
Q: Type in “[your name] looks like” in Google search.
A: Allen looks like he is barely holding it together.
Q: Type in “[your name] says” in Google search.
A: Allen says that multiple income residual programs are the way to make money using multiple income streams.
Q: Type in “[your name] wants” in Google search.
A: Allen wants shipbuilding excluded from trade agreement.
Q: Type in “[your name] does” in Google search.
A: Allen does not emphasize, though it is implicit in his analysis, the vulnerability of African-born and African-descended peoples in the early colonies.
(Editor’s note: What?)
Q: Type in “[your name] hates” in Google search.
A: Allen Hates Short Term Missions.
Q: Type in “[your name] asks” in Google search.
A: ALLEN ASKS MOSS TO FIND HER BOYFRIEND.
Q: Type in “[your name] goes” in Google search.
A: When Allen goes down, the Vikings rise up.
Q: Type in “[your name] likes ” in Google search.
A: ALLEN LIKES TO SMOKE
(Editor’s note: I don’t)
Q: Type in “[your name] eats ” in Google search.
A: Allen Eats Cabbage!
(Editor’s note: Again, I don’t. Ewwww!)
Q: Type in “[your name] wears ” in Google search.
A: Allen wears about 300000 thousand worth of jewelry.
Q: Type in “[your name] was arrested for” in Google Search.
A: Virginia’s warrant records make it look like Allen was arrested for not returning his textbooks in good to fair condition.
Why is it that projects like Gentoo only seem to think about projects that are needed / wanted when Google Summer of Code comes around? Surely this is something which should be ongoing!
Many of the projects on the Gentoo GSoC 2009 Ideas wiki page are probably things which devs have known needed doing for a while. But it’s only when GSoC comes around that these are put down in writing and start to become the formulation for a project.
I think that if this sort of thing happened on a continual basis, it might help to bring more developers to Gentoo and focus the efforts of both existing and potential developers / contributors.
I would suggest that, for Gentoo, the PR project maintains such a list on an ongoing basis (anyone would be able to edit the list, but PR would keep it tidy and ensure projects are updated (eg. removed when done)). The newsletter could feature one of these tasks each issue, with information from the task proposer on exactly what they forsee the project achieving.
There should be no upper or lower limits to the tasks. It could be as small as updating and testing a particular package or a major project (eg. resurrecting Gentoo/BSD).
A random find for today: The Starling Software Programmer Competency Matrix
An interesting method of evaluating programmer capabilities. Looking at this I can certainly see some areas I could easily work on to improve my ratings based on this system, but I don’t think I fair too badly overall.