On Wed, Sep 18, 2019 at 08:50:10PM +0200, Greg KH wrote: > > exfat is just a random old code drop from Samsung hastily picked up, > > and also duplicating the fat16/32 functionality, and without > > consultation of the developes of that who are trying to work through > > their process of contributing the uptodate and official version of it. > > Those developers are still yet to be found, we only got a "drop" of the > samsung code _after_ this code was merged, from non-samsung people. No > samsung developers have said they would actually help with any of this > work, and when asked what differed from the in-tree stuff, I got a list, > but no specifics. I'll be working through that list over the next month > to resolve those issues. Park offered to help with a new version, and the Samsung guys are going through their internal review process to work upstream it. Remember it is their codebase in each of the variants here. While it is fine if we take some code that has been abandoned by the original developers and still merge it with helping hands I think it is very rude to not at least wait for them to get their act of working with their corporate overlords together first. It isn't like we've been waiting forever for an exfat driver - the patent grant has been extremely recent, and this whole thing just seems like a publicity stunt to be honest. Give them a couple weeks to sort their stuff out and then we can decide how to proceed. I for one would defintively prefer to have code maintained by the original maintainers if possible. And not have them hindered by the staging process to work on their own code. > To get back to the meta-problem here, we need a common "vfs/filesystem" > tree somewhere with reviewers. I'm glad to set up the tree and handle > patches, but I am not a vfs expert by any means (I only understand the > virtual half, not the backing half), so I can't be the only one to do > reviews. Do you know of anyone that we can drag in as reviewers to help > make it easier for new filesystems to get reviews as well as existing > ones to have their patches collected and forwarded on to Linus at the > proper times? Following how staging works and its arcane rules I don't want it to be anywhere near fs code. And seriously, it is not like we need one magic tree to deal with all file systems. The whole point of git is that people can setup a tree to collaborate wherever they want if you just find people to work on it. And we've had tons of successful drivers and filesystems that worked that way. And at least the ones I've followed seem a lot more productive than the staging show that is baѕed around coding style cleanup micropatches. _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
