On Thursday 18 September 2008, Gadiyar, Anand wrote: > > (Hint for TI folk: why not write the > > code that way in the first place, avoiding all the delays inherent in > > writing code you *know* is unsuitable for merging to mainline?) > > Well, there are only a limited number of hours in a day. :) Bzip2 is a good way to compress things ... can that help make those hours more productive? ;) > Hint taken: that that code was initially written badly is a valid point. And likewise, I think, that it's not an uncommon way for things to start out. The real problem is that such code then sits in some (OMAP) tree for a long time and never gets fixed and sent upstream. One TI engineer (who I won't name!) said to me this year that he thinks the issue is part of TI's current process. Paraphrasing: the code is treated as "done" when it merges to some OMAP branch ... *NOT* when it goes upstream. To the extent that's true, it should get fixed. For drivers, there should be no problems doing so. (Infrastructure is often a bit different.) > But fixing it up is a non-trivial effort as is making sure it works. At the time I > wrote that e-mail (August 25) I did intend to have it out well before Felipe's > 11 September plan. > > As they say, the best laid plans of mice and men... At the moment it does > not look like I'll be able to do this anytime soon. Right. That happens. > And if it's going upstream, it needs to be in better shape than it currently is. > > Even assuming that linux-usb decides to take it in it's current shape, it's > definitely not going to help things when the OHCI driver comes in. And if > I'm going to be doing that cleanup anyway, I might as well get it right the > first time. Correct? Unless that stretches the EHCI merge out too much. Drivers sitting around outside of mainline for a long time (a year or more) make trouble. Updates to mainline generally don't make trouble. > I'm going to be re-arranging this code, writing it the way it ought to have > been the first time. But please don't expect this tomorrow. No, I'm not ... as I noted! Though I'd hope it's ready to push upstream in the 2.6.29-rc0 window... I'm just piling on with a few observations about a process problem here. We've seen versions of this for a long time now. Once you achieve that re-arrangement ... it'd be nicer all around (for you, for me, for everyone) if the problem came up less regularly in the future. - Dave -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html