Opened 14 years ago
Closed 13 years ago
#66 closed task (fixed)
Tests for DEFINE-METHOD-COMBINATION (long form)
| Reported by: | ehuelsmann | Owned by: | unassigned |
|---|---|---|---|
| Priority: | critical | Milestone: | 1.0 |
| Component: | CLOS | Version: | 1.0 |
| Keywords: | Cc: | ||
| Parent Tickets: |
Description (last modified by )
The issue used to be lack of support for D-M-C, now, it's about porting SBCL's tests (and required fixes) to ABCL; SBCL doesn't use RT - that's the main issue.
Attachments (2)
Change History (17)
comment:1 Changed 14 years ago by
| Milestone: | → 0.19 |
|---|---|
| Owner: | changed from somebody to Mark Evenson |
| Status: | new → assigned |
comment:2 Changed 14 years ago by
Another suggestion - by Tobias Rittweiler - is to copy the code from SACLA:
"
SACLA seems to contain an implementation of the long form of D-M-C:
http://homepage1.nifty.com/bmonkey/lisp/sacla/html/lisp/clos.lisp.html
It's released under a BSD license, so just copying it into ABCL is
viable, but the code may also just be used for inspiration.
"
comment:3 Changed 14 years ago by
| Milestone: | 0.19 → 0.20 |
|---|
comment:4 Changed 14 years ago by
| Priority: | major → blocker |
|---|
From my analysis of both codebases it seems like XCL did indeed copy the code from SACLA, so this is the preferred integration strategy.
comment:5 Changed 14 years ago by
| Owner: | changed from Mark Evenson to unassigned |
|---|
I am not actively working on this, so marking this as available in case someone else wants to charge ahead.
comment:6 Changed 14 years ago by
| Milestone: | 0.20 → 0.21 |
|---|---|
| Version: | → 1.0 |
Changed 14 years ago by
| Attachment: | run-d-m-c-20100802a.patch added |
|---|
Integration of SACLA/XCL DEFINE-METHOD-COMBINATION
comment:7 Changed 14 years ago by
| Milestone: | 0.21 → 0.22 |
|---|
A checkpoint of work that may be of interest to those who have more time than I currently do.
The attached patch takes a stab at integrating the work in Sacla/XCL into ABCL but does not work.
Arguments for long D-M-C seem to be correct and a new structure is in place, but calling this does not work as we now need an implementation of CALL-METHOD which should be a locally defined macro in the context of the method call.
Further analysis and help on how this would work is welcome
comment:8 Changed 14 years ago by
A further checkpoint that actually passes the long form of method combination in the ANSI test suite which is still a bit unbelievable.
Changed 14 years ago by
| Attachment: | run-d-m-c-20100811a.diff added |
|---|
long form of DEFINE-METHOD-COMBINATION snapshot
comment:9 Changed 14 years ago by
| Milestone: | 0.22 → 0.23 |
|---|
With Mark's pre-work, this should be doable in the next month.
comment:10 Changed 14 years ago by
| Priority: | blocker → critical |
|---|---|
| Type: | defect → task |
Current status:
Mark's patch is in trunk.
We need tests... Adjusting subject line accordingly
(from: Support for DEFINE-METHOD-COMBINATION (long form))
comment:11 Changed 14 years ago by
| Description: | modified (diff) |
|---|---|
| Summary: | Support for DEFINE-METHOD-COMBINATION (long form) → Tests for DEFINE-METHOD-COMBINATION (long form) |
comment:12 Changed 14 years ago by
As for being more descriptive about the non-functional parts of my patch, I think CALL-NEXT-METHOD does not currently do anything useful somehow failing to be expanded to the macro definition.
comment:13 Changed 13 years ago by
| Milestone: | 0.23 → unscheduled |
|---|
comment:14 Changed 13 years ago by
| Milestone: | unscheduled → 1.0 |
|---|
comment:15 Changed 13 years ago by
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
Fixed with r13593 (and earlier).
XCL which has extended and modified ABCL's CLOS has implemented the long form of DEFINE-METHOD-COMBINATION in XCL-0.0.0.290.
Backporting shouldn't be that much of a problem (once I understand what is going on).