-
Notifications
You must be signed in to change notification settings - Fork 8
Minutes
- Tom
- Brad
- Aury
- Magne
- Damian
- Tom
- Brad
- Damian
- Jon
- Mark
Eliminated remaining UTI's but found a new one because we edited some things in 2 places.
- Tom
- Brad ??? (did not keep notes)
Eliminated some UTIs
- Tom
- Aury
- Mark
- Jon
Worked on main syntax paper - incorporating comments from Tom and Brad created at end of J3 meeting.
6 UTI's were identified - mostly things that were too hard to do in the limited time.
- Tom
- Brad
- Aury
- Mark
- Magne
- Damian
Finished up paper on detached template procedures (now 106). Added syntax straw vote to simplified template procedures (now 105r1). Uploaded old main syntax paper (now 107).
- Tom
- Brad
- Lorri
- Aury
- Mark
Brad and Aury provided draft papers for requirements and specs on STPs and DTPs. In subgroup we further refined the papers. STP's seems done and DTP's is close. Am going to upload STP to J3.
- Tom
- Brad
- Lorri
- Magne
- Aury
- Mark
- Damian
We went over examples prepared by Aury showing simplified template syntax as well as some variations for full templates.
One major decision was to exclude any inference of template arguments based on actual arguments for now. This really only works when a template procedure only has type parameters, and the number of such use cases is actually rather small. (SWAP, REVERSE, TRANSPOSE, FINDLOC, etc.)
We generally like the idea of moving procedures provided by a regular template into the CONTAINS section of the host program unit. Solves the "long template" issue. Syntax for nested cases seems robust.
The issue of long argument lists was also discussed, but we are tabling that until after Feb meeting as we are behind.
Next week we should proceed with 2 info papers:
- Simplified syntax
- Moving template procedures to CONTAINS section of host program unit.
- Tom
- Brad
- Jon
- Magne
- Aury
- Mark
- Damian
A small amount of discussion about simplified templates, but most discussion was about whether additional constraints are necessary to reduce the complexity of processor analysis of source code. E.g., can INSTANTIATE reference procedures that are defined later in the module. Sort of went in a big circle as it was found (a) not to be as hard as originally thought and (b) desirable subcases (instantiation inside of module procedure) would create the pain anyway.
- Tom
- Brad
- Jon
- Magne
- Aury
- Mark
- Damian
- We have some general convergence on syntax to support simplified template use cases. Still some minor debate on low-level details (like which brackets to use for template args).
- Modified constraint on circular references in instantiation statements with procedure template args.
- Decided another constraint is needed for circular derived types in templates, but did not yet create the text.