Commits
- Commit:
6b36ff28e431165968013873f5f6ceac19d8aea4
- From:
- Omar Polo <op@omarpolo.com>
- Date:
change GotMArc::parse to take the line instead of $_
- Commit:
bbdbef1aa611793df3db28ba27fa2ef8bb34cfad
- From:
- Omar Polo <op@omarpolo.com>
- Date:
make GotMArc::parse return an hash reference instead of a list of strings
- Commit:
70b47196e34f4f26bce03cb299d75aee2b6e34f6
- From:
- Omar Polo <op@omarpolo.com>
- Date:
build a cache of hashes for each thread
The idea is to use this cache to spot changed threads and re-export
all the mail of the threads that changed. This will allow to add
thread navigation hints on each exported mail and have them up-to-date.
This makes the "skip if already exported" logic in mexp redundant
and incomplete. pe will decide whether to export or not, and mexp
blindly do the work.
- Commit:
75983083f49930cbff74c2bdbbce265dc5302009
- From:
- Omar Polo <op@omarpolo.com>
- Date:
pe: restrict pledge
the children are pre-forked, so at runtime only "stdio" is needed.
- Commit:
1bcb98996d9263f4db8045d0b71c8029749ca0bf
- From:
- Omar Polo <op@omarpolo.com>
- Date:
state the license on each file
- Commit:
de557185bc5f862ef773205dd8fb0f659bb4c1c2
- From:
- Omar Polo <op@omarpolo.com>
- Date:
refactor the parsing into the module
change the mscan format string to siplify the parsing, and refactor
the parsing code into the module for reuse.
pe is an exception in that it doesn't care about the format string, it
just need to decide if the current line starts a new thread or not,
that's why it doesn't use the newly introduced `parse'. The new
format simplifies pe too though.
- Commit:
707e3f4a96c2cec1375b104d687026440e1b7abe
- From:
- Omar Polo <op@omarpolo.com>
- Date:
pe: be less chatty
- Commit:
054f3fd48e662b95ddb20d25898866681971de6e
- From:
- Omar Polo <op@omarpolo.com>
- Date:
pledge and unveil the scripts
- Commit:
27363023e413886fbe0851562fd1f02be2467d98
- From:
- Omar Polo <op@omarpolo.com>
- Date:
parallelize exporting
mexp is the slowest part of the pipeline. Try to speed it up by
paralellizing it.
this adds a `pe' scripts that sits in front on N mexp childs and
dispatch threads to them, while still outputting the unchanged lines
for mkindex.
Unscientific testing has shown that this effectively reduces the time
for a full export, even if not linearly. (4 jobs cut the time in half,
8 jobs was just barely faster)