This is pure Go implementation of DJB's redo (http://cr.yp.to/redo.html)
build system proposal. Originally it was just a rewrite of redo-c
(https://github.com/leahneukirchen/redo-c), but later most features of
apenwarr/redo (https://redo.readthedocs.io/en/latest/) were also
implemented. Why yet another implementation? It is feature full and has
better performance comparing to shell and Python implementation.
goredo is free software: see the file COPYING for copying conditions.
Home page: http://www.goredo.cypherpunks.ru/
If you have problems with *.golang.org's unability to verify
authenticity of go.cypherpunks.ru TLS connection, then you can disable
their usage by setting GOPRIVATE=go.cypherpunks.ru. If you still have
problems with the authenticity on your side, then build it manually:
* "all" target is default
* stdout is always captured, but no target is created if it was empty
* empty targets are considered always out of date
* .do's $3 is relative path to the file in same directory
* .do search goes up to / by default, but can be limited with either
$REDO_TOP_DIR environment variable, or by having .redo/top file in it
* target's completion messages are written after they finish
* executable .do is run as is, non-executable is run with /bin/sh -e[x]
* tracing (-x) can be done only for non-executable .do
* explicit useful and convenient checks from apenwarr/redo:
* check that $1 was not touched during .do execution
* check that stdout and $3 are not written simultaneously
* check that generated target was not modified "externally" outside
the redo, preventing its overwriting, but continuing the build
* targets, dependency information and their directories are explicitly
synced (can be disabled, should work faster)
* file's change is detected by comparing its ctime and BLAKE2b hash
* files creation is umask-friendly (unlike mkstemp() used in redo-c)
* parallel build with jobs limit, optionally in infinite mode
* coloured messages (can be disabled)
* verbose debug messages, including out-of-date determination, PIDs,
lock and jobserver acquirings/releases
* displaying of each target's execution time
* each target's stderr can be prefixed with the PID
* optional statusline with currently running/waiting/done jobs
* target's stderr can be stored on the disk with TAI64N timestamp
prefixes for each line. To convert them to localtime you can use either
tai64nlocal utility from daemontools (http://cr.yp.to/daemontools.html)
or make a symlink, to use built-in slower decoder:
* redo-ifchange, redo-ifcreate, redo-always
* redo -- same as redo-ifchange, but forcefully and sequentially run
* redo-log -- display TAI64N timestamped last stderr of the target
* redo-stamp -- record stamp dependency. Nothing more, dummy
* redo-cleanup -- removes either temporary, log files, or everything
related to goredo
* redo-whichdo -- .do search paths for specified target (similar to
* redo-dot -- dependency DOT graph generator. For example to visualize
your dependencies with GraphViz:
Hashing and stamping
All targets are checksummed if their ctime differs from the previous
one. apenwarr/redo gives many reasons why every time checksumming is
bad, but in my opinion in practice all of them do not apply.
* Aggregate targets and willing to be out-of-date ones just must not
produce empty output files. apenwarr/*, redo-c and goredo
implementations consider non existing file as an out-of-date target
* If you really wish to produce an empty target file, just touch $3
DJB's proposal with both stdout and $3 gives that ability to control
your desired behaviour. Those who does not capture stdout -- failed.
Those who creates an empty file if no stdout was written -- failed.
redo is a tool to help people. Literally all targets can be safely
"redo-stamp < $3"-ed, reducing false positive out-of-dates. Of course,
with the correct stdout/$3 working and placing necessary results in $3,
instead of just silently feeding them in redo-stamp.
redo implementations are already automatically record -ifchange on .do
files and -ifcreate on non-existing .do files. So why they can not
record redo-stamp the same way implicitly? No, Zen of Python does not
applicable there, because -ifchange/-ifcreate contradict it already.
Modern cryptographic hash algorithms and CPUs are so fast, that even all
read and writes to or from hard drive arrays can be easily checksummed
and transparently compressed, as ZFS with LZ4 and Skein/BLAKE
algorithms demonstrate us.
goredo includes redo-stamp, that really records the stamp in the .dep
file, but it does not play any role later. It is stayed just for
Removed .do can lead to permanent errors of its non existence
That is true, because dependency on it was recorded previously. Is it
safe to assume that .do-less target now is an ordinary source-file? I
have no confidence in such behaviour. So it is user's decision how to
deal with it, probably it was just his inaccuracy mistake. If you really
want to get rid of that dependency knowledge for foo/bar target, then
just remove foo/.redo/bar.dep.
Dependency and build state is kept inside .redo subdirectory in each
directory related the build. Each corresponding target has its own,
recreated with every rebuild, .dep file. It is recfile
(https://www.gnu.org/software/recutils/), that could have various
dependency information (dep.rec with the schema included):
Run any of the command above with the -help.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, version 3 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.