Mache

Cache stdout of any process.

• The code is concise and written in Bash.
• Only caches stdout of processes that exit with status code 0, which avoids a bunch of edge cases.
• Will re-cache when the contents of the script changes or the options supplied to it change.
• Has a REMACHE variable which will refresh your caches.
• Supports printf style formatting.
• Define the cache file with MACHE_TARGET. Handy when doing builds.

Usage

Notice how it outputs the same date. It's using the cache.

mache date
sleep 2
mache date

za 25 mrt 2023 19:25:50 CET
za 25 mrt 2023 19:25:50 CET

If we supply different options it will create a new cache.

mache 'date -d "1 Apr 2020"'
mache 'date -d "2 Apr 2020"'

wo  1 apr 2020  0:00:00 CEST
do  2 apr 2020  0:00:00 CEST

You can also provide a script with arguments. This does not require one to quote the complete command.

mache /bin/echo hello
mache /bin/echo hello world

hello
hello world

Notice that it recomputes the cache for both commands separately. This is because the options are not equal.

Use printf style formatting.

mache 'echo "%s"' hello world
mache 'echo "%s"' 'hello world'

hello
world
hello world

You can also make a mache script. These scripts will cache by default. Simply define the following shebang at the top of that script.

cat ./t/date

#!/usr/bin/env mache
#!/usr/bin/env bash

date "$@"

We can now run that script assuming you made it executable. (chmod +x ./t/date)

./t/date

vr 24 mei 2024 14:13:28 CEST

By default mache will use the cache if present. You can remake the cache:

./t/date
sleep 2
REMACHE=1 ./t/date

vr 24 mei 2024 14:13:28 CEST
vr 24 mei 2024 14:13:38 CEST

A mache script does not cache when the script exits with non zero.

The arguments/options that are passed to the mache script are taken into account when checking if cache should be used. Different values will result in the mache script being re-evaluated. memoization

Tests

Requires bash-tap to be downloaded and on the $PATH.

Tests can be run with Perl's prove command; or execute a single one:

./t/printf.t # run single test

prove # run all tests

1..1
# 
ok - Supports printf like templating.
t/machetarget.t .. ok
t/printf.t ....... ok
t/shellcheck.t ... ok
All tests successful.
Files=3, Tests=5,  1 wallclock secs ( 0.02 usr  0.00 sys +  0.08 cusr  0.00 csys =  0.10 CPU)
Result: PASS

Caution

Caching can result in hard to find bugs because it isn't clear which version of the output is being used. When caching, cache thoughtfully and invalidate a cache when in doubt.

Documentation

Generating the documentation requires markatzea.

markatzea ./README.mz > ./README.md

Roadmap

• Use some sort of temporary filesystem for high read and write speeds. https://docs.oracle.com/cd/E19683-01/817-3814/6mjcp0r0l/index.html
• Consider having different cache for different STDIN.
• Consider having flock be separate from mache command.

License

GNU General Public License 3.0