Inline hash key_hash and compare functions

[luben]

Move runtime indirection to compile time indirection. This gives us some speedup.

For code review, look at branches/hash_inlined_func

[luben]

The work is done.

It's ready for testing, review and comments.

[jkeenan]

Failures observed in this branch in t/pmc/hash.t, most recently at r48871. For full output of prove, see  this paste. Excerpt:

src/hash.c:443: failed assertion 'key'
Backtrace - Obtained 16 stack frames (max trace depth is 32).
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e222c2]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_confess+0x9a) [0xb7e2242a]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e355d4]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(parrot_hash_put+0x1ba) [0xb7e360ca]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7ee3cb0]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7dea2ff]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e7c844]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e7b6ea]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e3db15]
/home/jimk/work/hash_inlined_func/blib/lib
  /libparrot.so.2.7.0(Parrot_pcc_invoke_from_sig_object+0x1c2) [0xb7e37a02]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_ext_call+0x103) [0xb7e23ef3]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_runcode+0x1b2) [0xb7e20d52]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(imcc_run_pbc+0x99) [0xb7f63bd9]
./parrot [0x8049a21]
/lib/libc.so.6(__libc_start_main+0xe5) [0xb70a3455]
./parrot [0x8048df1]
t/pmc/hash.t .. 
1..174
ok 1 - lookup Int in hash
...
ok 85 - clone preserves hash internal order
not ok 86 # TODO 
# 	Failed (TODO) test ''
ok 87 - Str from compound key
...
ok 152 - literal unicode key lookup via var
Failed 22/174 subtests 

Test Summary Report
-------------------
t/pmc/hash.t (Wstat: 6 Tests: 152 Failed: 0)
  Non-zero wait status: 6
  Parse errors: Bad plan.  You planned 174 tests but ran 152.
Files=1, Tests=152,  1 wallclock secs ( 0.03 usr  0.01 sys +  0.23 cusr  0.00 csys =  0.27 CPU)
Result: FAIL

[chromatic]

On Wednesday 08 September 2010 at 20:00, Parrot  wrote:

>  Failures observed in this branch in ''t/pmc/hash.t'', most recently at
>  r48871.  For full output of ''prove'', see [http://nopaste.snit.ch/23275
>  this paste].

r48879 should fix this.

-- c

[luben]

It was fixed by chromatic++ - argument annotations were mismatched.

[jkeenan]

Replying to luben:

It was fixed by chromatic++ - argument annotations were mismatched.

Unfortunately, no, for me, r48879 did not fix the problem (linux/i386):

$ svn info | ack 'Last.*Rev'
Last Changed Rev: 48879

$ prove t/pmc/hash.t 
t/pmc/hash.t .. 49/174 src/hash.c:476: failed assertion 'a'
Backtrace - Obtained 16 stack frames (max trace depth is 32).
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e292c2]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_confess+0x9a) [0xb7e2942a]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e3c2a4]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(parrot_hash_put+0x1fc) [0xb7e3d0dc]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7eec6b1]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7df0c66]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e83814]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e826ba]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0 [0xb7e44ae5]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_pcc_invoke_from_sig_object+0x1c2) [0xb7e3e9d2]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_ext_call+0x103) [0xb7e2aef3]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(Parrot_runcode+0x1b2) [0xb7e27d52]
/home/jimk/work/hash_inlined_func/blib/lib/libparrot.so.2.7.0(imcc_run_pbc+0x99) [0xb7f6aba9]
./parrot [0x8049a21]
/lib/libc.so.6(__libc_start_main+0xe5) [0xb70aa455]
./parrot [0x8048df1]
t/pmc/hash.t .. Failed 22/174 subtests 

Test Summary Report
-------------------
t/pmc/hash.t (Wstat: 6 Tests: 152 Failed: 0)
  Non-zero wait status: 6
  Parse errors: Bad plan.  You planned 174 tests but ran 152.
Files=1, Tests=152,  1 wallclock secs ( 0.04 usr  0.00 sys +  0.31 cusr  0.02 csys =  0.37 CPU)
Result: FAIL

[chromatic]

On Thursday 09 September 2010 at 19:33, Parrot  wrote:

>  Unfortunately, no, for me, r48879 did not fix the problem (linux/i386):
>  {{{
>  $ svn info | ack 'Last.*Rev'
>  Last Changed Rev: 48879
>  }}}
> 
>  {{{
>  $ prove t/pmc/hash.t
>  t/pmc/hash.t .. 49/174 src/hash.c:476: failed assertion 'a'
>  Backtrace - Obtained 16 stack frames (max trace depth is 32).

That's a different but similar problem which r48901 should fix.

-- c

[jkeenan]

Replying to chromatic:

That's a different but similar problem which r48901 should fix.

Enfin, ça va! t/pmc/hash.t PASS; now re-running make test.

kid51

[jkeenan]

r48922: continues to PASS make test Linux/i386.

[jkeenan]

The foregoing discussion refers to a hash_inlined_func branch, but there is no notation in the ticket that indicates if/when the branch was merged into trunk.

However, from study of git blame and git log, it appears that the relevant work was done by Luben Karavelov, Nick Wellnhofer and chromatic back in September and October of last year.

On this assumption, I'm going to take the ticket for the purpose of closing it in two or three days. Any objections or comments?

Thank you very much.

kid51

[jkeenan]

No comments or complaints; closing ticket.