Just getting this on record before the 1.0 release, I'll reveal it later.
5 plus ones
Shared publicly•View activity
View 22 previous comments
- Hmm, yes. However, Duncan mentioned that it will not be before the next GHC release. So you'll have to eagerly wait some more time... ;-)Apr 12, 2012
- Yes, text providing native builder support would/will also be nice.Apr 12, 2012
- I thought text already had a Builder in it...Apr 12, 2012
- Yes, it does have a builder. What is missing are functions like 'encodeUtf8Builder :: Text -> Builder' which can be used to assemble output from smallish Text values in a zero-copy fashion with a guarantee on a large average chunk size (required for few system calls).
Moreover, the new builder infrastructure also provides support for implementing fast custom encodings (mostly used for escaping + UTF-8 and the like). See this thread for some benchmarks: https://github.com/bos/aeson/issues/56#issuecomment-3834829
In the future, I imagine that functions creating sequences of bytes always result in builders instead of bytestrings, as this gives good compositionality + good performance. For this to happen, builders must be provided together with the bytestring library.Apr 13, 2012
- , it seems that the site is not working anymore. I've just remembered about this post and I'm not able to see what your prediction was =(...Nov 28, 2012
- Turns out I was totally wrong: http://www.whosawthatcoming.com/private/FEQNRBSXTS . I expected a lot of dynamic typing fans to pop in, I don't really remember seeing any. I based this prediction on the fact that I hear that claim made so often before I start demoing Yesod (once I give concrete examples, it's hard to argue against them).
The site's still working, but the domain name without the leading www is pointing at the wrong IP address. I'll get that fixed.Nov 29, 2012