We had a meeting of the Mercury group on Fri Dec 01, 3:15pm to 4:15pm. Attendees: David JefferyMark Brown Zoltan Somogyi Tyson Dowd Simon Taylor Fergus Henderson David Overton Levi Cameron Ina Cheng Julien Fischer Simon Mei Sam Thurairatnam Adrian Pellas-Rice (Record numbers!) 1. 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. Tyson Dowd - has spent a lot of time this week working with the summer students. - has been working on getting the IL backend library working. All the changes have been made, but there are some hitches with `:- external'. Tyson says he is still on track for a December release of this. The changes will be backported into the release branch after the code freeze. Mark Brown : - has been working on redesigning the search strategy for the declarative debugger. Simon Taylor - has been working on Aditi nondet joins and on calling Mercury code from Aditi. - still no estimated release date for Aditi, but will be released RSN. - Fergus asked Simon to post his current code for calling Mercury from Aditi since it may be helpful for Levi to look at for his planned work of calling Mercury from the bytecode debugger. David Jeffery : - Has been working on integrating a new constraint solver into HAL. Zoltan Somogyi : - has just finished his marking. Congratulations to him for that and to Adrian for getting the top mark in 313. Fergus Henderson : - has spent time with the summer students. - has been looking at compiling Mercury for the lcc and cc compilers with both the LLDS and MLDS back ends. Targetting lcc seems to be broken even though Peter Ross got it working a few months ago. It was decided to set up a machine to run its nightly tests using lcc to prevent future software rot. - has been looking at the new features of GCC: - found 6 bugs which he reported to the GCC maintainers. - GCC's new integrated CPP should be useful for reducing compilation times with the LLDS back end. - GCC's improved tailcall optimisation looks promising. Fergus suggested (to the GCC mailing list) adding an extension to GCC to tell it when it was safe to do a tailcall. Some of the GCC developers liked this idea and some didn't. - has added a check to the bootcheck scripts to make sure `MR_' prefixes occur wherever they should. David Overton : - finished marking exams on Tuesday - is looking into problems related to mode checking and types with user defined equality. Is planning to look at the options for what to do about this -- will need to be discussed, probably at the next meeting. - is working on updating and tidying up mtags. Levi Cameron : - Planning to work on adding bytecode queries to the Mercury debugger. Has started looking at Bert Thompson's work on the bytecode interpreter. Ina Cheng : - Planning to develop a SOAP interface for Mercury. Has been doing background reading and looking at the current Mercury CGI interface. Has developed some XML samples to be added to the repository. Julien Fischer : - Planning to work on the Java backend. Simon Mei : - Planning to work on a ``quickcheck''-style test generator. Sam Thurairatnam : - Planning to work on the pretty printer. Adrian Pellas-Rice - Planning to work on sequence quantification. 2. Discussions about the release There was a discussion about when we will be ready to freeze the compiler (i.e. create a release branch) and what needs to be done before that. - Tyson would like to get `:- external' working for the IL backend. Will commit all he has done (with non-working bits commented out) before the freeze and backport the working version into the release branch some time before the release. Zoltan wants the IL backend to be working for the release. - Zoltan has a couple of outstanding bugs, including one in the new LLDS code generator. - Fergus mentioned an outstanding bug with bootstrapping in grade asm_fast.gc.par. Cause of this bug is unknown. Current workaround is to use hlc.gc.par instead. - Ina would like to get her XML samples commited. It was decided we would aim to create the release branch late on Monday afternoon (4 Dec). Minutes taken by David Overton .