Open GoogleCodeExporter opened 9 years ago
Issue 18 has been merged into this issue.
Original comment by dbh@google.com
on 16 Nov 2010 at 6:27
Issue 17 has been merged into this issue.
Original comment by dbh@google.com
on 16 Nov 2010 at 6:27
Issue 7 has been merged into this issue.
Original comment by dbh@google.com
on 16 Nov 2010 at 6:30
szlbootstrapsum_unittest.cc, elfgen_unittest.cc and
intrinsics/resourcestats-32.szl should be fixed as of r41. Still looking at:
* intrinsics/resourcestats-64.szl
* emitter/tables_misc_good.szl
Original comment by dbh@google.com
on 16 Nov 2010 at 7:43
Some more failures on an amd64 Ubuntu natty chroot (szl built without NDEBUG).
In native mode, tracebacks have a bunch of extra frames.
Example: $ ./szl engine/language_tests/while_statements/bad_undef.szl >
/dev/null
* Golden output
Stack trace:
0. $main: function(input: bytes, input_key: bytes)
input: bytes = B""
input_key: bytes = B""
x: int = <undefined>
* Output on natty:
Stack trace:
0. $main: function(input: bytes, input_key: bytes)
input: bytes = B""
input_key: bytes = B""
x: int = <undefined>
1. no symbolic frame information (initialization frame?)
(use --v=1 for more details)
2. no symbolic frame information (initialization frame?)
(use --v=1 for more details)
3. no symbolic frame information (initialization frame?)
(use --v=1 for more details)
4. no symbolic frame information (initialization frame?)
(use --v=1 for more details)
Original comment by dbh@google.com
on 25 Nov 2010 at 6:56
I get an error in the regression test. Attached is the log of 'make check'. I
assume that most occurences of the word 'failed' are ok except for the one
where it mentions '*** FAILED ***'?
This happens on Ubuntu 10.10, all development headers from apt, szl trunk
checkout r50
Original comment by edwin.sm...@gmail.com
on 9 Dec 2010 at 3:55
Attachments:
Yep, the only failed test is:
emitter/tables_misc_good.szl
*** FAILED ***
It's the same problem mentioned above (caused by output sorting order with
different standard libraries). Here's part of the diff in your log:
88,89d87
< max2[{ A: B }] = { g: x }, 3
< max2[{ A: B, a: b }] = { : v, a: z, c: K }, 5
90a89,90
> max2[{ A: B, a: b }] = { : v, a: z, c: K }, 5
> max2[{ A: B }] = { g: x }, 3
97a98
> min1[] = {:}, 2
100d100
< min1[] = {:}, 2
102,103d101
< min2[{ A: B }] = { g: x }, 3
< min2[{ A: B, a: b }] = { : v, a: z, c: K }, 5
104a103,104
> min2[{ A: B, a: b }] = { : v, a: z, c: K }, 5
> min2[{ A: B }] = { g: x }, 3
124,126d123
...
Original comment by davi...@gmail.com
on 9 Dec 2010 at 4:17
Tests fail on MacOSX x86_64 because the regress script has a typo of "32" for
"64".
Patch provided.
Original comment by aecolley
on 13 Oct 2013 at 9:35
Attachments:
The locale-dependent sort order causes false failures of tables_misc_good.szl
(as mentioned in comment #7). Here's the fix.
Original comment by aecolley
on 13 Oct 2013 at 9:38
Attachments:
Original issue reported on code.google.com by
dbh@google.com
on 16 Nov 2010 at 6:26