1 *goredo* redo implementation on pure Go
3 OVERVIEW *goredo-overview*
5 This is pure Go implementation of DJB's redo (http://cr.yp.to/redo.html)
6 build system proposal. Originally it was just a rewrite of redo-c
7 (https://github.com/leahneukirchen/redo-c), but later most features of
8 apenwarr/redo (https://redo.readthedocs.io/en/latest/) were also
9 implemented. Why yet another implementation? It is feature full and has
10 better performance comparing to shell and Python implementation.
12 goredo is free software: see the file COPYING for copying conditions.
13 Home page: http://www.goredo.cypherpunks.ru/
15 INSTALL *goredo-install*
17 $ go get go.cypherpunks.ru/goredo
19 $ export PATH=`pwd`:$PATH
21 If you have problems with *.golang.org's unability to verify
22 authenticity of go.cypherpunks.ru TLS connection, then you can disable
23 their usage by setting GOPRIVATE=go.cypherpunks.ru. If you still have
24 problems with the authenticity on your side, then build it manually: >
26 $ git clone git://git.cypherpunks.ru/goredo.git
29 $ git clone git://git.cypherpunks.ru/gorecfile.git
30 $ ( cd gorecfile ; git tag -v v0.3.0 )
31 $ echo "replace go.cypherpunks.ru/recfile => `pwd`/gorecfile" >> go.mod
32 $ git clone git://git.cypherpunks.ru/gotai64n.git
33 $ ( cd gotai64n ; git tag -v v0.2.0 )
34 $ echo "replace go.cypherpunks.ru/tai64n => `pwd`/gotai64n" >> go.mod
37 $ export PATH=`pwd`:$PATH
41 * "all" target is default
42 * stdout is always captured, but no target is created if it was empty
43 * empty targets are considered always out of date
44 * .do's $3 is relative path to the file in same directory
45 * .do search goes up to / by default, but can be limited with either
46 $REDO_TOP_DIR environment variable, or by having .redo/top file in it
47 * target's completion messages are written after they finish
48 * executable .do is run as is, non-executable is run with /bin/sh -e[x]
49 * tracing (-x) can be done only for non-executable .do
51 FEATURES *goredo-features*
53 * explicit useful and convenient checks from apenwarr/redo:
54 * check that $1 was not touched during .do execution
55 * check that stdout and $3 are not written simultaneously
56 * check that generated target was not modified "externally" outside
57 the redo, preventing its overwriting, but continuing the build
58 * targets, dependency information and their directories are explicitly
59 synced (can be disabled, should work faster)
60 * file's change is detected by comparing its ctime and BLAKE2b hash
61 * files creation is umask-friendly (unlike mkstemp() used in redo-c)
62 * parallel build with jobs limit, optionally in infinite mode
63 * coloured messages (can be disabled)
64 * verbose debug messages, including out-of-date determination, PIDs,
65 lock and jobserver acquirings/releases
66 * displaying of each target's execution time
67 * each target's stderr can be prefixed with the PID
68 * optional statusline with currently running/waiting/done jobs
69 * target's stderr can be stored on the disk with TAI64N timestamp
70 prefixes for each line. To convert them to localtime you can use either
71 tai64nlocal utility from daemontools (http://cr.yp.to/daemontools.html),
73 $ go get go.cypherpunks.ru/tai64n/cmd/tai64nlocal
75 COMMANDS *goredo-commands*
77 * redo-ifchange, redo-ifcreate, redo-always
78 * redo -- same as redo-ifchange, but forcefully and sequentially run
80 * redo-log -- display TAI64N timestamped last stderr of the target
81 * redo-stamp -- record stamp dependency. Nothing more, dummy
82 * redo-cleanup -- removes either temporary, log files, or everything
84 * redo-whichdo -- .do search paths for specified target (similar to
86 $ redo-whichdo x/y/a.b.o
99 * redo-dot -- dependency DOT graph generator. For example to visualize
100 your dependencies with GraphViz: >
101 $ redo target [...] # to assure that **/.redo/*.dep are filled up
102 $ redo-dot target [...] > whatever.dot
103 $ dot -Tpng whatever.dot > whatever.png # possibly add -Gsplines=ortho
107 Hashing and stamping~
109 All targets are checksummed if their ctime differs from the previous
110 one. apenwarr/redo gives many reasons why every time checksumming is
111 bad, but in my opinion in practice all of them do not apply.
113 * Aggregate targets and willing to be out-of-date ones just must not
114 produce empty output files. apenwarr/*, redo-c and goredo
115 implementations consider non existing file as an out-of-date target
116 * If you really wish to produce an empty target file, just touch $3
118 DJB's proposal with both stdout and $3 gives that ability to control
119 your desired behaviour. Those who does not capture stdout -- failed.
120 Those who creates an empty file if no stdout was written -- failed.
122 redo is a tool to help people. Literally all targets can be safely
123 "redo-stamp < $3"-ed, reducing false positive out-of-dates. Of course,
124 with the correct stdout/$3 working and placing necessary results in $3,
125 instead of just silently feeding them in redo-stamp.
127 redo implementations are already automatically record -ifchange on .do
128 files and -ifcreate on non-existing .do files. So why they can not
129 record redo-stamp the same way implicitly? No, Zen of Python does not
130 applicable there, because -ifchange/-ifcreate contradict it already.
132 Modern cryptographic hash algorithms and CPUs are so fast, that even all
133 read and writes to or from hard drive arrays can be easily checksummed
134 and transparently compressed, as ZFS with LZ4 and Skein/BLAKE[23]
135 algorithms demonstrate us.
137 goredo includes redo-stamp, that really records the stamp in the .dep
138 file, but it does not play any role later. It is stayed just for
141 Can removed .do lead to permanent errors of its non existence?~
143 Yes, because dependency on it was recorded previously. Is it safe to
144 assume that .do-less target now is an ordinary source-file? I have no
145 confidence in such behaviour. So it is user's decision how to deal with
146 it, probably it was just his inaccuracy mistake. If you really want to
147 get rid of that dependency knowledge for foo/bar target, then just
148 remove foo/.redo/bar.dep.
150 Does redo-always always rebuilds target?~
152 goredo, together with apenwarr/redo, rebuilds target once per run.
153 Always rebuilds during every redo invocation, but only once during it
154 building. http://news.dieweltistgarnichtso.net/bin/redo-sh.html#why-built-twice
155 has other opinion, that is why its redo-sh.tests/always_rebuild_1 will
156 fail. Rebuilding of always-ed targets even during the same build process
157 ruins any redo's usability in practice.
159 For example if my .h file contains source code's version number, that is
160 git-describe's output and all my other files depends on that header,
161 then any redo-ifchange of .o will lead to git-describe execution, that
162 is rather heavy. Of course, because of either hashing or possible
163 redo-stamp-ing its dependants won't be rebuilt further, but build time
164 will be already ruined. If you need to rebuild TeX documents (case
165 mentioned in redo-sh's FAQ) until all references and numbers are ready,
166 then you must naturally expectedly explicitly use while cycle in your
167 .do, as apenwarr/redo already suggests.
171 Dependency and build state is kept inside .redo subdirectory in each
172 directory related the build. Each corresponding target has its own,
173 recreated with every rebuild, .dep file. It is recfile
174 (https://www.gnu.org/software/recutils/), that could have various
175 dependency information (dep.rec with the schema included): >
177 Build: 80143f04-bfff-4673-950c-081d712f573d
184 Ctime: 1605721341.253305000
185 Hash: f4929732f96f11e6d4ebe94536b5edef426d00ed0146853e37a87f4295e18eda
190 Hash: 5bbdf635932cb16b9127e69b6f3872577efed338f0a4ab6f2c7ca3df6ce50cc9
194 Run any of the command above with the -help.
198 This program is free software: you can redistribute it and/or modify
199 it under the terms of the GNU General Public License as published by
200 the Free Software Foundation, version 3 of the License.
202 This program is distributed in the hope that it will be useful,
203 but WITHOUT ANY WARRANTY; without even the implied warranty of
204 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
205 GNU General Public License for more details.