-
Notifications
You must be signed in to change notification settings - Fork 3
/
TODO
27 lines (18 loc) · 943 Bytes
/
TODO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
The current scheme of making candidates has N^2 memory usage. Keep
just the names of the candidates ("10", "10 semi") and generate the
candidates as they are needed.
write out a final file for each round, in case the user gets tired and
hits ^C then wants to restart
explore using Deparse -fFILE to expand package files. generally this
would be done after an initial run to minimize the test program.
worked-out example on a large codebase that isn't a single file
* ack is already combined up
make a tmp dir, put temp files there, put --keep files there
parallelize using Parallel::Fork or Parallel::Jobs or something.
* N temp dirs. Only need to communicated back win/lose via exitval
make end-to-end tests easier to run as actual tests
* fake (ack)
* real (dependent on perl version, alas... perlbrew instructions?)
** a system for expressing which tests should work/fail for a given perl version?
more unit tests
package for CPAN