… but it has taken almost 2 hours now to do a simple find and replace in a (admittedly 400k lines long) single file.
That is very inefficient.
In other news, if you tried to access this server this morning, you likely couldn’t. Why?
$ uptime
19:50:39 up 152 days, 3:40, 4 users, load average: 0.68, 31.71, 64.91
That is *not* what one wants to see. A runaway image resize daemon was to blame. Couldn’t even run ssh; had to drop to console over IP and kill the offending processes, then restart ssh et al.
I want some better hardware … and I really should reboot one of these months.
Update: I ended up killing TextMate and rethinking my methods.
Tags: textmate
November 1st, 2009 at 8:41 pm
This kind of pathological performance (loading entire files into memory, and keeping them there) is why I jumped ship.
The problem is bad even for small files; the key example being project-wide searches in things like Rails projects (very small files, but lots of them; all of them are kept in memory).
Vim, Emacs, any serious editor — even BBEdit, Xcode et al — handle these workloads easily, and that’s exactly what one would expect. The technology for handling this sensibly has existed for decades.
TextMate’s failure here is very disappointing, and the glacial pace of development of the new version doesn’t offer much hope (nor are there any guarantees that these bad design choices aren’t being carried forward into the new version).