We have a New Plan for the threaded RTS, which is stable, I think. Wolfgang Thaller is going to (a) write it up (or, rather, tidy up the existing design sketches) and (b) implement it in GHC (which is mostly done). But he's been busy recently. I'm sure that a chorus of hungry users would encourage him! The main thing the New Plan covers is the idea of a "bound thread", so that you can guarantee which OS thread will perform a foreign call; I think the lack of this guarantee is what breaks HOpenGL at the moment. The idea is that "threaded-rts" would be the default. I'm not sure it's worth making a threaded-rts variant distribution right now, given that we'd just throw it away later. But you're welcome to try. Wolfgang, do you have a timescale in mind? Simon | -----Original Message----- | From: glasgow-haskell-users-admin@haskell.org [mailto:glasgow-haskell-users-admin@haskell.org] | On Behalf Of Ian Lynagh | Sent: 22 July 2003 18:35 | To: glasgow-haskell-users@haskell.org | Subject: threaded-rts | | | Hi all, | | I've had a request for --enable-threaded-rts support in the Debian | packages, but I don't want to just build with it as then HOpenGL can't | be used. | | My hope was that I could share most files and just have a different | binary or something for the threaded-rts (it would be even better if a | threaded-rts enabled binary could have it disabled by a +RTS flag, but I | don't think this is possible?). | | However, various .hi, .a files and HSrts.o also seem to differ. So is | the best way forward to create two completely separate packages? | | | Thanks | Ian | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users@haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
I'm not sure it's worth making a threaded-rts variant distribution right now, given that we'd just throw it away later. But you're welcome to try.
We might not need the variant business when the new threaded-rts is finished, but personally, I like the thought of getting the "Threaded RTS classic" out there so it gets tested a little more. The new bound threads variant will share a lot of code with the classic variant, so testing the old code a little more does make sense.
Wolfgang, do you have a timescale in mind?
Well, let's see. I got my Powerbook back from repairs yesterday; now I have to track down one Mac-specific issue for 6.0.1. Then I'll have a look at the head and my old bound-threads prototypes. I might have some (preliminary, experimental, hackish, untested, ugly) patches ready on the weekend. Cheers, Wolfgang
On Wednesday, July 23, 2003, at 05:43 AM, Wolfgang Thaller wrote:
Well, let's see. I got my Powerbook back from repairs yesterday; now I have to track down one Mac-specific issue for 6.0.1. Then I'll have a look at the head and my old bound-threads prototypes. I might have some (preliminary, experimental, hackish, untested, ugly) patches ready on the weekend.
Cheers,
Wolfgang
Hi Wolfgang, I built the cvs version of 6.0.1 on Sunday (the 20th of July) and had the linker problem mentioned by Axel Simon (symbol _Main_zdmain_closure undefined). The build platform was OS X 10.2.6. If this isn't the OS X bug you're working on let me know. I should have a chance to look into it more deeply over the weekend. Otherwise, send a note when you have something you would like tested. I'm getting much better at building the ghc on my powerbook. ;-) Best Wishes, Greg
Hi Wolfgang, I had a good compile of 6.0.1 from Thursday's cvs on Mac OS X 10.2.6. I did a stage 2 build using your 6.0 package, then installed 6.0.1 and rebuilt everything. I ran but a few test programs (mostly some networking stuff I've been playing with); superficially it all works. A minor issue: ghc advertises itself as version 6.1, ghci as 6.0.1. Which is correct? Best Wishes, Greg
Hi Wolfgang,
I had a good compile of 6.0.1 from Thursday's cvs on Mac OS X 10.2.6. I did a stage 2 build using your 6.0 package, then installed 6.0.1 and rebuilt everything. I ran but a few test programs (mostly some networking stuff I've been playing with); superficially it all works.
Good... if anything acts strangely, just yell.
A minor issue: ghc advertises itself as version 6.1, ghci as 6.0.1. Which is correct?
When you say "6.01" from cvs, you are probably referring to the STABLE branch of the CVS. The stable branch will be released as "GHC 6.0.1" in a few days. My copy reports 6.0.1 throughout, where does it say 6.1? On the other hand, the HEAD branch of the CVS will be released as GHC 6.2 in a few months (?); in the meantime, it should advertise itself as GHC 6.1. Cheers, Wolfgang
When you say "6.01" from cvs, you are probably referring to the STABLE branch of the CVS. The stable branch will be released as "GHC 6.0.1" in a few days. My copy reports 6.0.1 throughout, where does it say 6.1?
On the other hand, the HEAD branch of the CVS will be released as GHC 6.2 in a few months (?); in the meantime, it should advertise itself as GHC 6.1.
This all makes sense now---I had built an earlier version off the HEAD, and another off the 6_0 branch. Both worked OK. The case of my powerbook is running at 35 C all of the time from all the building it's doing. Best Wishes, Greg
participants (3)
-
Gregory Wright -
Simon Peyton-Jones -
Wolfgang Thaller