I'm hoping a TM dev could point me in the right direction of where TM's inode caching takes place please?
I'm struggling to find it, but I think it might be one of the sources of my issues using TM on a mounted remote filesystem
Quite often, if I change git branch on the remote filesystem (done via ssh cmdline) it can end up with TM 'quick open' opening a completely different file. This issue persists until the volume is unmounted and remounted.
I'd like to compile a custom TM which disables inode caching if possible
Thanks
On 9 Feb 2017, at 4:16, Carpii UK wrote:
I'm hoping a TM dev could point me in the right direction of where TM's inode caching takes place please?
I'm struggling to find it, but I think it might be one of the sources of my issues using TM on a mounted remote filesystem
There isn’t any inode caching per se, however, when opening a file, TextMate will check if you already have that file open, and this check does involve looking at inodes.
So a possible explanation for your issue _could_ be that your virtual file system reuses inodes, and that you open a file that uses an inode of something you previously opened (but which had a different path), and is still open somehow, e.g. in a background tab.
Perhaps you can ask the creators of the virtual file system that you are using, how they deal with inodes, i.e. if there is a high chance of inode re-use. If they say no, my theory is invalid, but if they say yes, then perhaps there is something to it.
Thanks Allan
It's not a virtual filesystem I'm using, so much as a mounted volume (smb, running on vmware esxi)
I upgraded to Sierra a couple of weeks ago, and Ive just realised I've not noticed any TM issues since then. I'm hopeful all of the issues I've been facing recently (atomic save conflicts, wrong files opening from Quick Open etc) were due to samba incompatibilities or bugs in my earlier OSX
Thanks Paul
On 5 March 2017 at 04:05, Allan Odgaard mailinglist@textmate.org wrote:
On 9 Feb 2017, at 4:16, Carpii UK wrote:
I'm hoping a TM dev could point me in the right direction of where TM's inode caching takes place please?
I'm struggling to find it, but I think it might be one of the sources of my issues using TM on a mounted remote filesystem
There isn’t any inode caching per se, however, when opening a file, TextMate will check if you already have that file open, and this check does involve looking at inodes.
So a possible explanation for your issue *could* be that your virtual file system reuses inodes, and that you open a file that uses an inode of something you previously opened (but which had a different path), and is still open somehow, e.g. in a background tab.
Perhaps you can ask the creators of the virtual file system that you are using, how they deal with inodes, i.e. if there is a high chance of inode re-use. If they say no, my theory is invalid, but if they say yes, then perhaps there is something to it.
textmate mailing list textmate@lists.macromates.com http://lists.macromates.com/listinfo/textmate
-
Allan Odgaard -
Carpii UK