| commit | 58b6a3d2d379d39c531686b42f329c46ac5ee81b | |
| author | rofl0r <retnyg@gmx.net> | |
| Wed, 8 Jun 2016 17:14:59 +0000 (8 18:14 +0100) | ||
| committer | rofl0r <retnyg@gmx.net> | |
| Wed, 8 Jun 2016 17:15:03 +0000 (8 18:15 +0100) | ||
| tree | 9ff2ae7bba801073dd0d7a29afd03811d6e10ed8 | treesnapshot (tar.gz zip) |
| parent | a4942ceaa6f50e4d8ebb5bf7eb0b5bb685e0679a | commitdiff |
remove confusing usage of slot list
while reading the code, i found the usage of the slot list rather
confusing. the intent was to optimize looking up of a free job slot
(index into the job array), instead of iterating over the job array
to find an unused entry (pid == -1). however this can be entirely
left away if we just fill the job array sequentially till it's full,
and later by waiting for/reaping a single process and re-use its
job slot immediately when we want to add a job.
while reading the code, i found the usage of the slot list rather
confusing. the intent was to optimize looking up of a free job slot
(index into the job array), instead of iterating over the job array
to find an unused entry (pid == -1). however this can be entirely
left away if we just fill the job array sequentially till it's full,
and later by waiting for/reaping a single process and re-use its
job slot immediately when we want to add a job.
| jobflow.c | diffblobblamehistory |