On Tue, 28 Jun 2011 22:40:00 +0200 Yann Dirson <dirson@debian.org> wrote: > Hi Neil, > > First, let me thank you for this detailed emdebian status. Maybe a > link to it could be added to the website (eg. a "Status" link under > "About emdebian"). Things change a bit too quickly for that to stay up to date. I try to use the mailing list for all such status messages. (See my recent post about MultiArch.) However, some things (like the thing with Grip) are not entirely in the public domain yet in terms of the details of implementation, so have to be only referred to indirectly (which is where this thread started). > > Issues around trimming back multistrap to something which is actually > > cohesive are completely unrelated to the processing of Grip on buildds. > > I can understand that, but work in the rootfs-production area, being > orthogonal as you mentionned, can be done by other people too. > Activity around polystrap is a good illustration, with various people > working on this subject. I may be wrong, but the work on a qemu wrapper around multistrap is probably a result of a discussion between Wookey and myself (IIRC on the train to FOSDEM) where I put my foot down and said 'Non' to adding yet more layers to multistrap itself. ;-) In order to keep things simple and create small tools which do a small number of tasks very well, it is sometimes necessary to upset people and refuse to allow particular changes. This gets one a reputation for being dismissive, rude and/or discouraging - especially when there is insufficient time to explain the reasons again and again. It also isn't a comfortable situation for a developer - the temptation to just add a small extra feature is hard to resist but it just leads to another request for another small extra feature. I finally decided to refuse when I had already put in a few too many little extra features and now the door really is shut and it's time to sort out what's left behind. Some stuff will simply have to be removed and if people want it, they'll have to do what people did with polystrap - create another small tool which does that job well by using other small tools which do their own job well. > Yourself focusing on the Grip processing > issue is good thing, and you could delegate some multistrap > maintainance to someone else if you feel it will help you keep this > focus. It's not that I don't use multistrap on Squeeze - Squeeze is my daily development platform. I just don't think it is right to go fiddling with minor details now when much larger issues - which will still affect multistrap and need documenting - are imminent and when the whole issue of what should be in multistrap (and therefore documented) and what should not is up for grabs. > Not everyone has the same interest in both areas, and not everyone has > the necessary time to dedicate to the Grip issue; add this to the fact > that while work is being done on wheezy, those tools can still be > worked on by people who are content with squeeze. They will just be > more ready when Wheezy Grip get live! There are issues around that - updates for multistrap for Squeeze behaviour cannot be uploaded to Debian because that will be into unstable where MultiArch is being implemented. See my other (lengthy) email about exactly what's involved in that process. > For my part, my immediate interest is precisely rootfs generation for > a squeeze-based distro, and I don't have the time to go on the moving > ground that MultiArch currently is - which probably explains why we > were not understanding each other. > > I'll do my best to keep those issues separate :) Thank you. I'm glad that there can be a change of tone in our correspondence. I don't want to discourage people but I also object to behaviour which discourages me from my own activities. I try to keep the mailing list updated with currently pressing issues, wherever such things can be described publicly. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
Attachment:
pgpAauGxyDaRD.pgp
Description: PGP signature