On Tue, 2002-05-21 at 21:58, Seth Arnold wrote: > On Tue, May 21, 2002 at 09:11:24PM +0200, petter wahlman wrote: > > I am working in a hetrougenous development envoronment that share code > > through a source revision constrol system (not CVS). This is causing > > some problems (make, etc.) when e.g a source file has an include > > statement with incorrect case. This can easily be fixed, but it is not > > convinient when having to correct many files. > > Ah, every time I hear this, I recall a statement by one of the Unix or C > luminaries, describing whey they couldn't fix an acknowledged problem in > Unix or C. (I wish I could recall more specifically. Citations > welcomed. :) The quote goes something along the lines of, "We couldn't > make a change like that; there were already a dozen installations with > over 300 kilobytes of source code." > > > So, why are 'linux' filesystems case sensitive? > > Because Linux is modeled after Unix system, which for roughly twenty > years before the first version of Linux was released used case sensitive > filesystems. All the users expected case sensitive filesystems. Agreed. This is mostly a matter of personal preference, but the option to make the FS case insensitive would certainly be a usefull feature - I do not think that I am the only one experiensing such problems. > > > I have briefly looked at the VFS code, and are uncertain of what parts > > to change. Is it sufficient to only change fs/dcache: d_lookup? > > My guess is that this isn't likely. I bet you will need to change a > _lot_ of code, filesystem-specific and in the VFS layer. I'd be curious > to hear how it goes, but I don't expect it is within the range of most > organization's budgets. Hmm..I had hoped that changing the VFS would be sufficient, but you might be right. Making sure d_name is lowercase would probably work, but I want the FS to be able to use uppercase letters aswell. Sadly I do not have any organizations to fund such a change, and not as much spare time as I'd like. Any voluntaries :-) > > Actually, thinking about it, I bet the MSDOS FAT filesystem already > implemented case-insensitive code. You could just use FAT and be done > with it. VFAT code might be case insensitive too, so you might still get > long filenames out of the deal. Indeed. VFAT has a dentry operation: d_compare = vfat_cmpi wich can be activated on mount (check=relaxed). Argh, maybe I should say bye, bye to my journaling FS and degrade to VFAT. This actually seems like the best sollution. Thanks. > > I'd be interested in hearing how it goes. I'll give some feedback on the d_lookup change when I reboot (no UML or vmware on this machine yet) > > -- > New GPG key coming soon, please grab D9B0A099 before this one expires. -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
