On Tue, Aug 10, 2004 at 04:29:47PM -0400, Silbermann, Martine wrote: > Hi Greg, > > I'm glad that you brought up all those points and I don't pretend to > have answers > to all of them but I'll try to explain why we thought that creating a > SIG might > help the community as opposed to "clogging" it with another mailing > list. <offtopic> Please fix your email client to not wrap lines in such a weird way. I've reformatted the lines below to make it easier to read. </offtopic> > >> Why? Seriously, why create > yet-another-mailing-list-that-contains-no-real-work? > > The SIG's intention is to do real work and the first thing to do is to > determine what (if any) that would be to help the community. One of > the things I was thinking of organizing are "working meetings" where > some ongoing discussions could be hashed out in a "live phone meeting" > as opposed to long e-mail threads. OSDL has the infrastructure to > support such calls. Phone calls stink for technical discussions like this. Email gives everyone the means to express issues in detail, and it provides the time to give thoughtful replies to these issues. It also works very well for people for which English is not their native language. Providing a phone call doesn't really work when the developers are in Germany, India, China, Japan, and the US if for no other reason than the time differences. > Also SIGs are not created to hang around for ever, the goal is to figure > out what still needs to be done, write the code for it, get it > submitted and accepted and disband. Ok, but why create something to do this, when the groups are already out there doing this very work? > As you well know I have not been active in the Hotplug area and will > need to first understand what the needs are and if the SIG can help > along. > > For example I understand that most of the hotplug CPU code is written > but it also seems that there hasn't been much testing done. Is that > true? if yes why not? lack of equipment? I have not heard this complaint at all. There is code present to test CPU hotplug on machines that can not electrically remove CPUs for the reason that finding such hardware is quite difficult. In your testing (or others) have you found issues that point to a lack of testing of the hotplug CPU code? > For Hotplug IO subsystem, you mentioned > >> "They said they will be posting code soon (no real rush, as there > >> is no hardware shipping with this support yet.)". > A lot of OSDL members have access to such hardware long before it hits > the market, wouldn't it be helpful to test the code on real hw before > submitting it? All such code will be tested on such hardware before submitting it to the relevant kernel maintainer. The developers doing the work already have access to such hardware. Also, OSDL can not provide access to pre-production hardware for members of the community, so I do not see how they will be able to help out in this area. > >> So, what is this group going to provide the Linux community that > >> the others don't already cover? Why is it worth wasting people's > >> time listening to a conference call discussing the fact that all of > >> this work is already being done by other people, in other > >> discussion areas, in the open? > > I think THAT will be a key point i.e. to find out what/if/how the SIG > can help the community. I agree. But shouldn't have that been thought of before the SIG was formed? > I agree with you that there is no need to duplicate ongoing effort but > I'm not convinced (yet?!? :-) that there isn't stuff that needs to be > done for Hotplug that's still up for grabs. Great, if you find such stuff, please let us (the developers on the various lists) know about it. This really sounds like the Intel "hotplug" proposal (which I already discussed in public on the dcl list) all rehashed again. > Also as I mentioned earlier the SIG might be used just to promote > other forms of discussions other than the traditional mailing lists. > For ex. I'm sure that the BOFs at OLS were very helpful, would it be > feasible to organize a day (or more) of face to face discussions? Again, see my point above about language and distance issues. > I have a lot of homework to do but you already helped me greatly with > your mail. I have an idea that I'll regret asking this but is there > anything that you can think of that would complete or help along > ongoing work? Don't create new meetings or mailing lists that cause people who are doing the work to have spend time monitoring. Go to the developers, they aren't hiding, don't require them to come to you. Also try to determine what the real goals for this SIG are, and document it somewhere. If there's going to be something that will cause this SIG to be disbanded, please document that too (like "full support for all tasks in the mainline kernel.org kernel"). Good luck, greg k-h
