DMD 1.036 and 2.020
Big news in D-land today. Walter has pushed out two new DMD release, one each of the 1.x and 2.x branches. The big news is what's inside of them.
The 2.020 release got all of the attention in the newsgroup thread and no one has yet commented on the first item in the 1.036 changelog:
improved performance of AAs by rebalancing trees when rehashing
OK, that's not just a 1.x thing, as 2.020 sports the same fix. But it's a reason to give D1 some love!
The big, big news that has everyone jabbering is in the changelog for 2.020 and was the primary focus of Walter's announcement, in which he dropped this much-anticipated line:
The 2.0 version splits phobos into druntime and phobos libraries (thanks to Sean Kelly). This will enable both Tango and Phobos to share a common core library.
How's that for BIG NEWS? This is very bleeding edge stuff, so I suggest you read the entire announcement thread if you are going to play around with it. I've not touched it yet myself, but some workarounds for a few issues are discussed in NG. I'm not sure if Walter has updated the zip yet with the corrections(or if he even will). If you do play around and have any issues with druntime, be sure to post a ticket at the project site so as to keep Sean busy.
If you are a D noob, be aware that the 1.x series is 'stable' while the 2.x series, despite the 2.x versioning, is not considered stable or production ready. Let that guide you in deciding which version to jump into.
I'm rather happy about these releases. This wipes away a couple of the issues I had with D and finally gives me a reason to start looking at a new branch for Derelict (as if I'm not busy enough already). Stay tuned.
Technorati Tags: D Programming Language, DMD, Digital Mars, Walter Bright
October 21st, 2008 - 11:59
Re: no one commenting on “improved performance of AAs by rebalancing trees when rehashing”
Bearophile did, and he said he would benchmark them. It’s easy to accidentally de-optimize some important use cases while making others fast with AAs, so I’m waiting for the verdict from Bearophiles benchmarks till I pass judgement on that one.
But all the stuff for D2.0 is great. And it looks like now that D2 has reference returns Andrei is all set to unleash his Ranges and a new Range-based std.algorithm on the world within a release or two.