James Antill - The myth of the "simple String API", that isn't really needed
Oct. 4th, 2007
01:00 am - The myth of the "simple String API", that isn't really needed
I saw three things recently which just confirmed that noone learns anything:
- The "we don't need a real string API" on the GIT mailing list, followed by almost a month now of struggling to implement another one (and wishing they had some of the ustr features).
- GLibc's proposed fix for C/POSIX not having a usable string API. Yes, it's the old let's all pretend IO (FILE *) is a good string abstraction ... you need to add, search/etc, and then add some more? Well multiple copies were always good fun, pretty much guaranteeing people will do a bunch of workarounds if they use it at all.
- The Linux kernel is again looking for yet another almost a worthwhile string API, pretending it'll only be useful for this specific IO type ... because seq_printf() wasn't unique enough.
Best quote by far, IMO (from the GIT thread):
"Please, no. Let's not pull in a dependency for something as simple as a string library." -- Kristian Høgsberg
As a minor good note, ustr is in Fedora and so pretty much everyone in Fedora land has access to a usable string API with a single -l ... and the newer versions of libsemanage (SELinux) use ustr. I'm not going to hold my breath for mass sanity to occur though.