We had a meeting of the Mercury group on Fri Jan 19, 2001, from 3:15 4:05pm. Attendees: Fergus Henderson (fjh) David Jeffery (dgj) Julien Fischer (juliensf) Ina Cheng (inch) Sam Thurairatnam (sthur) Levi Cameron (lpcam) Zoltan Somogyi (zs) Adrian Pellas-Rice (apell) Simon Mei (xcsm) Tyson Dowd (trd) David Overton (dmo) Mark Brown (dougl) Simon Taylor (stayl) 1. Release - There are two outstanding bugs that need to be fixed before the release (in the the hlc backend, and one in asm_fast). These bugs have been recorded on sourceforge (as have other release tasks). - Tyson also wants to add a couple of fixes to the IL back end. These are non-critical (ie. release will go ahead without them, but they'll be included if there is time). 2. Progress reports We went around the table, with each person in turn discussing what they had been working on recently and/or what they were planning to work on. Fergus Henderson (fjh): - Continued working on the new gcc backend. He has been working on bug fixes, and integrating the new backend into mmake and the rest of the build environment. He has not yet tried bootstrapping or running the tests, but it works for everything in the samples and samples/diff directories (but not for all the other subdirectories of samples). - Has helped the summer students a bit. David Jeffery (dgj) - Has been working on HAL. - Has worked briefly with Fergus on a bug reported by petdr which resulted in a compiler abort in polymorphism. The cause is now clear, and a fix should come soon. Julien Fischer (juliensf) - Has been working on discriminated unions for the Java backend. He got enumerations working earlier in the week. - Fergus encouraged him to post a diff soon so he can get some decent feedback before the end of the summer. Ina Cheng (inch) - Has been working on encoding Mercury types in XML. Sam Thurairatnam (sthur) - Spent the week working on addressing dougl's review comments. Levi Cameron (lpcam) - The bytecode interpreter is working for simple deterministic code. - Has added support for handling modules to the interpreter. - Is working on getting the compiler to generate the stub functions that are currently written by hand. - Fergus encouraged him to post a diff of the bytecode interpreter in the next week. (It might be worth posting the whole interpreter too if the diff is too large to read). Zoltan Somogyi (zs) - Has worked on minimal model tabling with Kostis. The system now passes all of our test cases (although they aren't a very complete set). Kostis is going to send him a whole lot more test cases from XSB which should test things more thoroughly. - Has been working with Tom on deep profiling. Tom will run some benchmarks on the different mechanisms that they've implemented. This should be enough to finish off their paper. - Has found a problem with bogus output from the liveness pass causing havoc for the delay death transformation. - Needs to work on the bug in var.m for the release. Adrian Pellas-Rice (apell) - Has worked on implementing a simple version of sequence quantification in the compiler. Simon Mei (xcsm) - Added support for user defined generators to quickcheck. - fjh asked about how useful Simon thought this tool would be, and how it compared to the Haskell version, and encouraged him to think about this and maybe write something about it. Tyson Dowd (trd) - Fixed some IL backend bugs - Has added a new `developer intro' page to the www site - Has implemented floating point for the IL backend. - Has worked on implementing `foreign_proc'. - Has hacked fjh's `cvdr' script to handle branches. He should add this script to the `tools' directory. David Overton (dmo) - Has been working on the new mode checker. Mark Brown (dougl) - Has been tracking down a bug caused by Tom's changes to the `univ' implementation. (There was a bug in `compare' for univs). - Has been tracking down another bug caused by the fact that I/O now works differently for univ (as it is no longer handled specially). Tyson agreed to work on fixing io__read and io__write so that they work properly for existentially quantified types. Simon Taylor (stayl) - Still working on Aditi/Mercury integration, and working towards an Aditi release some time in the future. Minutes taken by David Jeffery (dgj@cs.mu.oz.au).