On Sat, Aug 14, 2010 at 5:11 PM, John Millikin <jmillikin@gmail.com> wrote:

This attitude towards performance, that it doesn't really matter as
long as something happens *eventually*, is what pushed me away from
Python and towards more performant languages like Haskell in the first
place.

But wait, wait - I'm not at all contending that performance doesn't matter! In fact, I spent a couple of months working on criterion precisely because I want to base my own performance work on extremely solid data, and to afford the same opportunity to other people. So far in this thread, there's been exactly one performance number posted, by Daniel. Not only have I already thanked him for it, I immediately used (and continue to use) it to improve the performance of the text library in that instance.

More broadly, what I am recommending is simple:

Use a good library.
Expect good performance out of it.
Measure the performance you get out of your application.
If it's not good enough, and the fault lies in a library you chose, report a bug and provide a test case.

In the case of the text library, it is often (but not always) competitive with bytestring, and I improve it when I can, especially when given test cases. My goal is for it to be the obvious choice on several fronts:

Cleanliness of API, where it's already better, but could still improve
Performance, which is not quite where I want it (target: parity with, or better than, bytestring)
Quality, where text has slightly more test coverage than bytestring

However, just text alone is a big project, and I could get a lot more done if I was both coding and integrating patches than if coding alone :-) So patches are very welcome.

> In the unlikely event that you need to support non-Unicode encodings,
> they are readily available via text-icu.

Unfortunately, text-icu is hardcoded to use libicu 4.0, which was
released well over a year ago and is no longer available in many
distributions. I sent you a patch to support newer versions a few
months ago, but never received a response.

Yes, that's quite embarrassing, and I am quite apologetic about it, especially since I just asked for help in the preceding paragraph. If it's any help, there's a story behind my apparent sloth: I overenthusiastically accepted a patch from another contributor a few months before yours, and his changes left the text-icu darcs repo in a mess from which I have yet to rescue it. I do still have your patch, and I'll probably abandon my attempts to clean up the other one, as it was more work than I cared to clean it up.