"make quickheap-CMD" generates a heap profile for CMD, in
hledgerprof.ps, and tries to open it in a viewer (currently the
mac-friendly "open" executable, so you may need to adjust this in the
makefile).
As with quickprof, CMD must be one word and runs against one of the
sample journals.
A bunch of cleanups to make profiling possible again, and easier.
"make quickprof-CMD" generates a profile for CMD, which must be one
word and runs against one of the sample journals (the usual
quickprof-"SOME WORDS" quoting trick isn't working here for some
reason)
Also,
"make hledgerprof" builds the hledgerprof executable (in stack's bin dir) for profiling.
"make hledgercov" builds the hledgercov executable (in ./bin) for coverage reports
(renamed from hledgerhpc to remind me that HPC is nothing to do with heap coverage)
Here are hpack package.yaml files for the other hledger cabal files.
These remove a lot of human-error-prone duplication.
They are not used yet as hpack isn't quite mature enough -
when it supports flags and benchmarks we will probably switch.
hakyll-std's cabal file is now generated by hpack from an easier yaml
config file, which looks like a valuable timesaver. "make gencabalfiles"
will regenerate this cabal file (and soon the others) when needed.
Drop deprecated shakespeare-text dependency for ghc 7.10+ consistently,
not just in one place. Hopefully this will clear the warning
at packdeps.haskellers.com.
The makefile now uses stack primarily.
The following updated rules are available:
$ make
Makefile:35: -------------------- hledger make rules --------------------
Makefile:37: make [help] -- list documented rules in this makefile. make -n RULE shows more detail.
Makefile:202: (INSTALLING:)
Makefile:204: make install -- download dependencies and install hledger executables to ~/.local/bin or equivalent (with stack)
Makefile:229: (BUILDING:)
Makefile:233: make build -- download dependencies and build hledger executables (with stack)
Makefile:302: make hledgerdev -- quickly build the hledger executable (with ghc and -DDEVELOPMENT)
Makefile:487: (TESTING:)
Makefile:495: make test -- run default tests
Makefile:541: make pkgtest -- run the test suites for each package
Makefile:548: make builtintest -- run tests built in to the hledger executable (subset of pkg tests)
Makefile:572: make functest -- run hledger functional tests with the stack build
Makefile:606: make haddocktest -- run haddock and make sure it succeeds
Makefile:787: (DOCUMENTATION:)
Makefile:895: make haddock -- generate haddock docs for the hledger packages
Makefile:974: (RELEASING:)
Makefile:1053: make setversion -- update all version strings to match .version
Makefile:1056: make setversionforce -- update all version strings even if .version seems unchanged
Makefile:1080: make tagrelease -- commit a release tag based on .version for each package
Makefile:1253: (MISCELLANEOUS:)
Makefile:1255: make usage -- show size of various dirs
Makefile:1260: make stackusage -- show size of stack working dirs
Makefile:1264: make cabalusage -- show size of cabal working dirs if any
Makefile:1268: make tag -- generate tag files for source code navigation (for emacs)
Makefile:1294: make clean -- default cleanup (ghc build leftovers)
Makefile:1297: make Clean -- thorough cleanup (stack/cabal/ghc builds and tags)
Makefile:1302: make cabalCMD -- run cabal CMD inside each hledger package directory
Makefile:1306: make all"CMD" -- run CMD inside each hledger package directory
Using "hledgerdev" was a hack to help ensure that tests used a fresh
developer build by default. Now they specify "hledger" again, which fits
better with stack. It's up to the tester to make sure the desired
executable is first in PATH or specified with -w. (Note a couple of
tests currently don't obey -w and will always run "hledger", see addons.test).
We don't do a good job of calculating good-looking unit prices when the
commodity display precisions are low. Eg when a journal doesn't use any
decimal places, any inferred unit prices are shown by the print command
also with no decimal places, which makes them look wrong.
Now inferred unit prices always have a minimum display precision of 2,
which helps a bit. Could do better.