OILS / build / deps.sh View on Github | oilshell.org

1134 lines, 579 significant
1#!/usr/bin/env bash
2#
3# Script for contributors to build dev dependencies -- packaged as cross-distro
4# "wedges". Tested in the Soil CI.
5#
6# Usage:
7# build/deps.sh <function name>
8#
9# Examples:
10# build/deps.sh fetch
11# build/deps.sh install-wedges-fast # for both Python and C++
12#
13# build/deps.sh rm-oils-crap # rm -r -f /wedge ~/wedge to start over
14#
15# TODO: Do we need something faster, just python2, re2c, and cmark?
16#
17# - build/deps.sh fetch-py
18# - build/deps.sh install-wedges-py
19#
20# TODO: Can we make most of them non-root deps? This requires rebuilding
21# containers, which requires podman.
22#
23# rm -r -f ~/wedge # would be better
24
25: ${LIB_OSH=stdlib/osh}
26source $LIB_OSH/bash-strict.sh
27source $LIB_OSH/task-five.sh
28
29REPO_ROOT=$(cd "$(dirname $0)/.."; pwd)
30
31source build/dev-shell.sh # python3 in PATH, PY3_LIBS_VERSION
32source deps/from-apt.sh # PY3_BUILD_DEPS
33#source deps/podman.sh
34source test/tsv-lib.sh # tsv-concat
35source web/table/html.sh # table-sort-{begin,end}
36
37# Also in build/dev-shell.sh
38USER_WEDGE_DIR=~/wedge/oils-for-unix.org
39ROOT_WEDGE_DIR=/wedge/oils-for-unix.org
40
41readonly DEPS_SOURCE_DIR=_build/deps-source
42
43readonly RE2C_VERSION=3.0
44readonly RE2C_URL="https://github.com/skvadrik/re2c/releases/download/$RE2C_VERSION/re2c-$RE2C_VERSION.tar.xz"
45
46readonly CMARK_VERSION=0.29.0
47readonly CMARK_URL="https://github.com/commonmark/cmark/archive/$CMARK_VERSION.tar.gz"
48
49readonly PY_FTP_MIRROR="${PY_FTP_MIRROR:-https://www.python.org/ftp}"
50
51readonly PY2_VERSION=2.7.18
52readonly PY2_URL="$PY_FTP_MIRROR/python/$PY2_VERSION/Python-$PY2_VERSION.tar.xz"
53
54readonly PY3_VERSION=3.10.4
55readonly PY3_URL="$PY_FTP_MIRROR/python/$PY3_VERSION/Python-$PY3_VERSION.tar.xz"
56
57readonly BASH_VER=4.4 # don't clobber BASH_VERSION
58readonly BASH_URL="https://www.oilshell.org/blob/spec-bin/bash-$BASH_VER.tar.gz"
59
60# Another version of bash to test
61readonly BASH5_VER=5.2.21
62readonly BASH5_URL="https://www.oilshell.org/blob/spec-bin/bash-$BASH5_VER.tar.gz"
63
64readonly DASH_VERSION=0.5.10.2
65readonly DASH_URL="https://www.oilshell.org/blob/spec-bin/dash-$DASH_VERSION.tar.gz"
66
67readonly ZSH_VERSION=5.1.1
68readonly ZSH_URL="https://www.oilshell.org/blob/spec-bin/zsh-$ZSH_VERSION.tar.xz"
69
70readonly MKSH_VERSION=R52c
71readonly MKSH_URL="https://www.oilshell.org/blob/spec-bin/mksh-$MKSH_VERSION.tgz"
72
73readonly BUSYBOX_VERSION='1.35.0'
74readonly BUSYBOX_URL="https://www.oilshell.org/blob/spec-bin/busybox-$BUSYBOX_VERSION.tar.bz2"
75
76readonly YASH_VERSION=2.49
77readonly YASH_URL="https://www.oilshell.org/blob/spec-bin/yash-$YASH_VERSION.tar.xz"
78
79readonly MYPY_GIT_URL=https://github.com/python/mypy
80readonly MYPY_VERSION=0.780
81
82readonly PY3_LIBS=~/wedge/oils-for-unix.org/pkg/py3-libs/$MYPY_VERSION
83
84# Version 2.4.0 from 2021-10-06 was the last version that supported Python 2
85# https://github.com/PyCQA/pyflakes/blob/main/NEWS.rst
86readonly PYFLAKES_VERSION=2.4.0
87#readonly PYFLAKES_URL='https://files.pythonhosted.org/packages/15/60/c577e54518086e98470e9088278247f4af1d39cb43bcbd731e2c307acd6a/pyflakes-2.4.0.tar.gz'
88# 2023-07: Mirrored to avoid network problem on broome during release
89readonly PYFLAKES_URL='https://www.oilshell.org/blob/pyflakes-2.4.0.tar.gz'
90
91readonly BLOATY_VERSION=1.1
92readonly BLOATY_URL='https://github.com/google/bloaty/releases/download/v1.1/bloaty-1.1.tar.bz2'
93
94readonly UFTRACE_VERSION=0.13
95readonly UFTRACE_URL='https://github.com/namhyung/uftrace/archive/refs/tags/v0.13.tar.gz'
96
97readonly SOUFFLE_VERSION=2.4.1
98readonly SOUFFLE_URL=https://github.com/souffle-lang/souffle/archive/refs/tags/2.4.1.tar.gz
99
100log() {
101 echo "$@" >& 2
102}
103
104die() {
105 log "$0: fatal: $@"
106 exit 1
107}
108
109rm-oils-crap() {
110 ### When you want to start over
111
112 rm -r -f -v ~/wedge
113 sudo rm -r -f -v /wedge
114}
115
116# Note: git is an implicit dependency -- that's how we got the repo in the
117# first place!
118
119# python2-dev is no longer available on Debian 12
120# python-dev also seems gone
121#
122# wget: for fetching wedges (not on Debian by default!)
123# tree: tiny package that's useful for showing what we installed
124# g++: essential
125# libreadline-dev: needed for the build/prepare.sh Python build.
126# gawk: used by spec-runner.sh for the special match() function.
127# cmake: for cmark
128# PY3_BUILD_DEPS - I think these will be used for building the Python 2 wedge
129# as well
130readonly -a WEDGE_DEPS_DEBIAN=(
131 bzip2
132 wget
133 tree
134 gawk
135 g++
136 ninja-build
137 cmake
138 libreadline-dev
139 systemtap-sdt-dev
140
141 # for Souffle, flex and bison
142 #flex bison
143
144 "${PY3_BUILD_DEPS[@]}"
145)
146
147readonly -a WEDGE_DEPS_ALPINE=(
148 bzip2
149 xz
150
151 wget tree gawk
152
153 gcc g++
154 ninja-build
155 # https://pkgs.alpinelinux.org/packages?name=ninja-is-really-ninja&branch=v3.19&repo=&arch=&maintainer=
156 ninja-is-really-ninja
157 cmake
158
159 readline-dev
160 zlib-dev
161 libffi-dev
162 openssl-dev
163
164 ncurses-dev
165
166 # for Souffle, flex and bison
167 #flex bison
168)
169
170readonly -a WEDGE_DEPS_FEDORA=(
171
172 # Weird, Fedora doesn't have these by default!
173 hostname
174 tar
175 bzip2
176
177 # https://packages.fedoraproject.org/pkgs/wget/wget/
178 wget
179 # https://packages.fedoraproject.org/pkgs/tree-pkg/tree/
180 tree
181 gawk
182
183 # https://packages.fedoraproject.org/pkgs/gcc/gcc/
184 gcc gcc-c++
185
186 ninja-build
187 cmake
188
189 readline-devel
190
191 # Like PY3_BUILD_DEPS
192 # https://packages.fedoraproject.org/pkgs/zlib/zlib-devel/
193 zlib-devel
194 # https://packages.fedoraproject.org/pkgs/libffi/libffi-devel/
195 libffi-devel
196 # https://packages.fedoraproject.org/pkgs/openssl/openssl-devel/
197 openssl-devel
198
199 # For building zsh from source?
200 # https://koji.fedoraproject.org/koji/rpminfo?rpmID=36987813
201 ncurses-devel
202 #libcap-devel
203
204 # still have a job control error compiling bash
205 # https://packages.fedoraproject.org/pkgs/glibc/glibc-devel/
206 # glibc-devel
207)
208
209install-debian-packages() {
210 ### Packages for build/py.sh all, building wedges, etc.
211
212 set -x # show what needs sudo
213
214 # pass -y for say gitpod
215 sudo apt "$@" install "${WEDGE_DEPS_DEBIAN[@]}"
216 set +x
217
218 # maybe pass -y through
219 test/spec-bin.sh install-shells-with-apt "$@"
220}
221
222install-ubuntu-packages() {
223 ### Debian and Ubuntu packages are the same; this function is suggested on the wiki
224 install-debian-packages "$@"
225}
226
227wedge-deps-debian() {
228 # Install packages without prompt
229
230 # 2024-02 - there was an Ubuntu update, and we started needing this
231 sudo apt-get -y update
232
233 install-debian-packages -y
234}
235
236wedge-deps-fedora() {
237 # https://linuxconfig.org/install-development-tools-on-redhat-8
238 # Trying to get past compile errors
239 # sudo dnf group install --assumeyes 'Development Tools'
240
241 sudo dnf install --assumeyes "${WEDGE_DEPS_FEDORA[@]}"
242}
243
244wedge-deps-alpine() {
245 # https://linuxconfig.org/install-development-tools-on-redhat-8
246 # Trying to get past compile errors
247 # sudo dnf group install --assumeyes 'Development Tools'
248
249 sudo apk add "${WEDGE_DEPS_ALPINE[@]}"
250}
251
252#
253# Unused patch, was experiment for Fedora
254#
255
256get-typed-ast-patch() {
257 curl -o deps/typed_ast.patch https://github.com/python/typed_ast/commit/123286721923ae8f3885dbfbad94d6ca940d5c96.patch
258}
259
260# Work around typed_ast bug:
261# https://github.com/python/typed_ast/issues/169
262#
263# Apply this patch
264# https://github.com/python/typed_ast/commit/123286721923ae8f3885dbfbad94d6ca940d5c96
265#
266# typed_ast is tarred up though
267patch-typed-ast() {
268 local package_dir=_cache/py3-libs
269 local patch=$PWD/deps/typed_ast.patch
270
271 pushd $package_dir
272 cat $patch
273 echo
274
275 local dir=typed_ast-1.4.3
276 local tar=typed_ast-1.4.3.tar.gz
277
278 echo OLD
279 ls -l $tar
280 echo
281
282 rm -r -f -v $dir
283 tar -x -z < $tar
284
285 pushd $dir
286 patch -p1 < $patch
287 popd
288 #find $dir
289
290 # Create a new one
291 tar --create --gzip --file $tar typed_ast-1.4.3
292
293 echo NEW
294 ls -l $tar
295 echo
296
297 popd
298}
299
300#
301# Fetch
302#
303
304download-to() {
305 local dir=$1
306 local url=$2
307 wget --no-clobber --directory-prefix "$dir" "$url"
308}
309
310maybe-extract() {
311 local wedge_dir=$1
312 local tar_name=$2
313 local out_dir=$3
314
315 if test -d "$wedge_dir/$out_dir"; then
316 log "Not extracting because $wedge_dir/$out_dir exists"
317 return
318 fi
319
320 local tar=$wedge_dir/$tar_name
321 case $tar_name in
322 *.gz|*.tgz) # mksh ends with .tgz
323 flag='--gzip'
324 ;;
325 *.bz2)
326 flag='--bzip2'
327 ;;
328 *.xz)
329 flag='--xz'
330 ;;
331 *)
332 die "tar with unknown extension: $tar_name"
333 ;;
334 esac
335
336 tar --extract $flag --file $tar --directory $wedge_dir
337}
338
339clone-mypy() {
340 ### replaces deps/from-git
341 local dest_dir=$1
342 local version=${2:-$MYPY_VERSION}
343
344 local dest=$dest_dir/mypy-$version
345 if test -d $dest; then
346 log "Not cloning because $dest exists"
347 return
348 fi
349
350 # v$VERSION is a tag, not a branch
351
352 # size optimization: --depth=1 --shallow-submodules
353 # https://git-scm.com/docs/git-clone
354
355 git clone --recursive --branch v$version \
356 --depth=1 --shallow-submodules \
357 $MYPY_GIT_URL $dest
358
359 # TODO: verify commit checksum
360}
361
362copy-source-medo() {
363 mkdir -p $DEPS_SOURCE_DIR
364
365 # Copy the whole tree, including the .treeptr files
366 cp --verbose --recursive --no-target-directory \
367 deps/source.medo/ $DEPS_SOURCE_DIR/
368}
369
370fetch-spec-bin() {
371 download-to $DEPS_SOURCE_DIR/bash "$BASH_URL"
372 maybe-extract $DEPS_SOURCE_DIR/bash "$(basename $BASH_URL)" bash-$BASH_VER
373
374 download-to $DEPS_SOURCE_DIR/bash "$BASH5_URL"
375 maybe-extract $DEPS_SOURCE_DIR/bash "$(basename $BASH5_URL)" bash-$BASH5_VER
376
377 download-to $DEPS_SOURCE_DIR/dash "$DASH_URL"
378 maybe-extract $DEPS_SOURCE_DIR/dash "$(basename $DASH_URL)" dash-$DASH_VERSION
379
380 download-to $DEPS_SOURCE_DIR/zsh "$ZSH_URL"
381 maybe-extract $DEPS_SOURCE_DIR/zsh "$(basename $ZSH_URL)" zsh-$ZSH_VERSION
382
383 download-to $DEPS_SOURCE_DIR/mksh "$MKSH_URL"
384 maybe-extract $DEPS_SOURCE_DIR/mksh "$(basename $MKSH_URL)" mksh-$MKSH_VERSION
385
386 download-to $DEPS_SOURCE_DIR/busybox "$BUSYBOX_URL"
387 maybe-extract $DEPS_SOURCE_DIR/busybox "$(basename $BUSYBOX_URL)" busybox-$BUSYBOX_VERSION
388
389 download-to $DEPS_SOURCE_DIR/yash "$YASH_URL"
390 maybe-extract $DEPS_SOURCE_DIR/yash "$(basename $YASH_URL)" yash-$YASH_VERSION
391
392 # Patch: this tarball doesn't follow the convention $name-$version
393 if test -d $DEPS_SOURCE_DIR/mksh/mksh; then
394 pushd $DEPS_SOURCE_DIR/mksh
395 mv -v mksh mksh-$MKSH_VERSION
396 popd
397 fi
398}
399
400fetch() {
401 local py_only=${1:-}
402
403 # For now, simulate what 'medo expand deps/source.medo _build/deps-source'
404 # would do: fetch compressed tarballs designated by .treeptr files, and
405 # expand them.
406
407 # _build/deps-source/
408 # re2c/
409 # WEDGE
410 # re2c-3.0/ # expanded .tar.xz file
411
412 copy-source-medo
413
414 download-to $DEPS_SOURCE_DIR/re2c "$RE2C_URL"
415 download-to $DEPS_SOURCE_DIR/cmark "$CMARK_URL"
416 maybe-extract $DEPS_SOURCE_DIR/re2c "$(basename $RE2C_URL)" re2c-$RE2C_VERSION
417 maybe-extract $DEPS_SOURCE_DIR/cmark "$(basename $CMARK_URL)" cmark-$CMARK_VERSION
418
419 if test -n "$py_only"; then
420 log "Fetched dependencies for 'build/py.sh'"
421 return
422 fi
423
424 download-to $DEPS_SOURCE_DIR/pyflakes "$PYFLAKES_URL"
425 maybe-extract $DEPS_SOURCE_DIR/pyflakes "$(basename $PYFLAKES_URL)" \
426 pyflakes-$PYFLAKES_VERSION
427
428 download-to $DEPS_SOURCE_DIR/python2 "$PY2_URL"
429 download-to $DEPS_SOURCE_DIR/python3 "$PY3_URL"
430 maybe-extract $DEPS_SOURCE_DIR/python2 "$(basename $PY2_URL)" Python-$PY2_VERSION
431 maybe-extract $DEPS_SOURCE_DIR/python3 "$(basename $PY3_URL)" Python-$PY3_VERSION
432
433 fetch-spec-bin
434
435 # bloaty and uftrace are for benchmarks, in containers
436 download-to $DEPS_SOURCE_DIR/bloaty "$BLOATY_URL"
437 download-to $DEPS_SOURCE_DIR/uftrace "$UFTRACE_URL"
438 maybe-extract $DEPS_SOURCE_DIR/bloaty "$(basename $BLOATY_URL)" bloaty-$BLOATY_VERSION
439 maybe-extract $DEPS_SOURCE_DIR/uftrace "$(basename $UFTRACE_URL)" uftrace-$UFTRACE_VERSION
440
441 # This is in $DEPS_SOURCE_DIR to COPY into containers, which mycpp will directly import.
442
443 # It's also copied into a wedge in install-wedges.
444 clone-mypy $DEPS_SOURCE_DIR/mypy
445
446 if false; then
447 download-to $DEPS_SOURCE_DIR/souffle "$SOUFFLE_URL"
448 maybe-extract $DEPS_SOURCE_DIR/souffle "$(basename $SOUFFLE_URL)" souffle-$SOUFFLE_VERSION
449 fi
450
451 if command -v tree > /dev/null; then
452 tree -L 2 $DEPS_SOURCE_DIR
453 fi
454}
455
456fetch-py() {
457 fetch py_only
458}
459
460mirror-pyflakes() {
461 ### Workaround for network error during release
462 scp \
463 $DEPS_SOURCE_DIR/pyflakes/"$(basename $PYFLAKES_URL)" \
464 oilshell.org:oilshell.org/blob/
465}
466
467mirror-python() {
468 ### Can't reach python.org from some machines
469 scp \
470 $DEPS_SOURCE_DIR/python2/"$(basename $PY2_URL)" \
471 oilshell.org:oilshell.org/blob/
472
473 scp \
474 $DEPS_SOURCE_DIR/python3/"$(basename $PY3_URL)" \
475 oilshell.org:oilshell.org/blob/
476}
477
478wedge-exists() {
479 ### Does an installed wedge already exist?
480
481 local name=$1
482 local version=$2
483 local wedge_dir=${3:-/wedge/oils-for-unix.org}
484
485 local installed=$wedge_dir/pkg/$name/$version
486
487 if test -d $installed; then
488 log "$installed already exists"
489 return 0
490 else
491 return 1
492 fi
493}
494
495#
496# Install
497#
498
499# TODO: py3-libs needs to be a WEDGE, so that that you can run
500# 'wedge build deps/source.medo/py3-libs/' and then get it in
501#
502# _build/wedge/{absolute,relative} # which one?
503#
504# It needs a BUILD DEPENDENCY on:
505# - the python3 wedge, so you can do python3 -m pip install.
506# - the mypy repo, which has test-requirements.txt
507
508download-py3-libs() {
509 ### Download source/binary packages, AFTER python3 is installed
510
511 # Note that this is NOT source code; there is binary code, e.g. in
512 # lxml-*.whl
513
514 local mypy_dir=${1:-$DEPS_SOURCE_DIR/mypy/mypy-$MYPY_VERSION}
515 local py_package_dir=_cache/py3-libs
516 mkdir -p $py_package_dir
517
518 # Avoids a warning, but doesn't fix typed_ast
519 #python3 -m pip download -d $py_package_dir wheel
520
521 python3 -m pip download -d $py_package_dir -r $mypy_dir/test-requirements.txt
522 python3 -m pip download -d $py_package_dir pexpect
523}
524
525install-py3-libs-in-venv() {
526 local venv_dir=$1
527 local mypy_dir=$2 # This is a param for host build vs. container build
528 local package_dir=_cache/py3-libs
529
530 source $venv_dir/bin/activate # enter virtualenv
531
532 # 2023-07 note: we're installing yapf in a DIFFERENT venv, because it
533 # conflicts with MyPy deps!
534 # "ERROR: pip's dependency resolver does not currently take into account all
535 # the packages that are installed."
536
537 # --find-links uses a "cache dir" for packages (weird flag name!)
538
539 # Avoids a warning, but doesn't fix typed_ast
540 #time python3 -m pip install --find-links $package_dir wheel
541
542 upgrade-typed-ast $mypy_dir/mypy-requirements.txt
543
544 # for mycpp/
545 time python3 -m pip install --find-links $package_dir -r $mypy_dir/test-requirements.txt
546
547 # pexpect: for spec/stateful/*.py
548 time python3 -m pip install --find-links $package_dir pexpect
549}
550
551upgrade-typed-ast() {
552 local file=$1
553 sed -i 's/typed_ast.*/typed_ast==1.5.0/' $file
554}
555
556test-typed-ast() {
557 local dir=~/wedge/oils-for-unix.org/pkg/mypy/0.780
558
559 cp -v $dir/mypy-requirements.txt _tmp
560
561 local file=_tmp/mypy-requirements.txt
562 cat $file
563 #echo
564
565 # 1.5.0 fixed this bug
566 # https://github.com/python/typed_ast/issues/169
567
568 upgrade-typed-ast $file
569 echo
570 cat $file
571}
572
573install-py3-libs-from-cache() {
574
575 # As well as end users
576
577 local mypy_dir=${1:-$DEPS_SOURCE_DIR/mypy/mypy-$MYPY_VERSION}
578
579 local py3
580 py3=$(command -v python3)
581 case $py3 in
582 *wedge/oils-for-unix.org/*)
583 ;;
584 *)
585 die "python3 is '$py3', but expected it to be in a wedge"
586 ;;
587 esac
588
589 log "Ensuring pip is installed (interpreter $(command -v python3)"
590 python3 -m ensurepip
591
592 local venv_dir=$USER_WEDGE_DIR/pkg/py3-libs/$PY3_LIBS_VERSION
593 log "Creating venv in $venv_dir"
594
595 # Note: the bin/python3 in this venv is a symlink to python3 in $PATH, i.e.
596 # the /wedge we just built
597 python3 -m venv $venv_dir
598
599 log "Installing MyPy deps in venv"
600
601 # Run in a subshell because it mutates shell state
602 $0 install-py3-libs-in-venv $venv_dir $mypy_dir
603}
604
605install-py3-libs() {
606 ### Invoked by Dockerfile.cpp-small, etc.
607
608 download-py3-libs
609 install-py3-libs-from-cache
610}
611
612
613# zsh notes
614 # Fedora compiler error
615 # zsh ./configure is NOT detecting 'boolcodes', and then it has a broken
616 # fallback in Src/Modules/termcap.c that causes a compile error! It seems
617 # like ncurses-devel should fix this, but it doesn't
618 #
619 # https://koji.fedoraproject.org/koji/rpminfo?rpmID=36987813
620 #
621 # from /home/build/oil/_build/deps-source/zsh/zsh-5.1.1/Src/Modules/termcap.c:38:
622 # /usr/include/term.h:783:56: note: previous declaration of ‘boolcodes’ with type ‘const char * const[]’
623 # 783 | extern NCURSES_EXPORT_VAR(NCURSES_CONST char * const ) boolcodes[];
624 #
625 # I think the ./configure is out of sync with the actual build?
626
627
628# TODO:
629# - $ROOT_WEDGE_DIR vs. $USER_WEDGE_DIR is duplicating information that's
630# already in each WEDGE file
631
632py-wedges() {
633 ### for build/py.sh all
634
635 echo cmark $CMARK_VERSION $ROOT_WEDGE_DIR
636 echo re2c $RE2C_VERSION $ROOT_WEDGE_DIR
637 echo python2 $PY2_VERSION $ROOT_WEDGE_DIR
638 echo pyflakes $PYFLAKES_VERSION $USER_WEDGE_DIR
639}
640
641cpp-wedges() {
642 ### for ninja / mycpp translation
643
644 echo python3 $PY3_VERSION $ROOT_WEDGE_DIR
645 echo mypy $MYPY_VERSION $USER_WEDGE_DIR
646
647 # py3-libs has a built time dep on both python3 and MyPy, so we're doing it
648 # separately for now
649 #echo py3-libs $PY3_LIBS_VERSION $USER_WEDGE_DIR
650}
651
652spec-bin-wedges() {
653 ### for test/spec-py.sh osh-all
654
655 echo dash $DASH_VERSION $USER_WEDGE_DIR
656 echo bash $BASH_VER $USER_WEDGE_DIR
657 echo bash $BASH5_VER $USER_WEDGE_DIR
658 echo mksh $MKSH_VERSION $USER_WEDGE_DIR
659 echo zsh $ZSH_VERSION $USER_WEDGE_DIR
660 echo busybox $BUSYBOX_VERSION $USER_WEDGE_DIR
661 echo yash $YASH_VERSION $USER_WEDGE_DIR
662}
663
664contributor-wedges() {
665 py-wedges
666 cpp-wedges
667 spec-bin-wedges
668}
669
670extra-wedges() {
671 # Contributors don't need uftrace, bloaty, and probably R-libs
672 # Although R-libs could be useful for benchmarks
673
674 # Test both outside the contianer, as well as inside?
675 echo uftrace $UFTRACE_VERSION $ROOT_WEDGE_DIR
676
677 #echo souffle $SOUFFLE_VERSION $USER_WEDGE_DIR
678}
679
680timestamp() {
681 date '+%H:%M:%S'
682}
683
684my-time-tsv() {
685 python3 benchmarks/time_.py \
686 --tsv \
687 --time-span --rusage \
688 "$@"
689}
690
691maybe-install-wedge() {
692 local name=$1
693 local version=$2
694 local wedge_dir=$3 # e.g. $USER_WEDGE_DIR or empty
695
696 local task_file=$WEDGE_LOG_DIR/$name-$version.task.tsv
697 local log_file=$WEDGE_LOG_DIR/$name-$version.log.txt
698
699 echo " TASK $(timestamp) $name $version > $log_file"
700
701 # python3 because it's OUTSIDE the container
702 # Separate columns that could be joined: number of files, total size
703 my-time-tsv --print-header \
704 --field xargs_slot \
705 --field wedge \
706 --field wedge_HREF \
707 --field version \
708 --output $task_file
709
710 if wedge-exists "$name" "$version" "$wedge_dir"; then
711 echo "CACHED $(timestamp) $name $version"
712 return
713 fi
714
715 local -a cmd=( deps/wedge.sh unboxed _build/deps-source/$name/ $version)
716
717 set +o errexit
718 my-time-tsv \
719 --field "$XARGS_SLOT" \
720 --field "$name" \
721 --field "$name-$version.log.txt" \
722 --field "$version" \
723 --append \
724 --output $task_file \
725 "${cmd[@]}" "$@" >$log_file 2>&1
726 local status=$?
727 set -o errexit
728
729 if test "$status" -eq 0; then
730 echo " OK $(timestamp) $name $version"
731 else
732 echo " FAIL $(timestamp) $name $version"
733 fi
734}
735
736dummy-task() {
737 ### For testing log capture
738 local name=$1
739 local version=$2
740
741 echo "Building $name $version"
742
743 # random float between 0 and 3
744 # weirdly we need a seed from bash
745 # https://stackoverflow.com/questions/4048378/random-numbers-generation-with-awk-in-bash-shell
746 local secs
747 secs=$(awk -v seed=$RANDOM 'END { srand(seed); print rand() * 3 }' < /dev/null)
748
749 echo "sleep $secs"
750 sleep $secs
751
752 echo 'stdout'
753 log 'stderr'
754
755 if test $name = 'mksh'; then
756 echo "simulate failure for $name"
757 exit 2
758 fi
759}
760
761readonly WEDGE_LOG_DIR=_build/wedge/logs
762
763dummy-task-wrapper() {
764 # Similar to test/common.sh run-task-with-status, used by
765 # test/{spec,wild}-runner.sh
766 local name=$1
767 local version=$2
768
769 local task_file=$WEDGE_LOG_DIR/$name.task.tsv
770 local log_file=$WEDGE_LOG_DIR/$name.log.txt
771
772 echo " TASK $(timestamp) $name $version > $log_file"
773
774 # python3 because it's OUTSIDE the container
775 # Separate columns that could be joined: number of files, total size
776 my-time-tsv --print-header \
777 --field xargs_slot \
778 --field wedge \
779 --field wedge_HREF \
780 --field version \
781 --output $task_file
782
783 my-time-tsv \
784 --field "$XARGS_SLOT" \
785 --field "$name" \
786 --field "$name.log.txt" \
787 --field "$version" \
788 --append \
789 --output $task_file \
790 $0 dummy-task "$@" >$log_file 2>&1 || true
791
792 echo " DONE $(timestamp) $name $version"
793}
794
795html-head() {
796 # python3 because we're outside containers
797 PYTHONPATH=. python3 doctools/html_head.py "$@"
798}
799
800index-html() {
801 local tasks_tsv=$1
802
803 local base_url='../../../web'
804 html-head --title 'Wedge Builds' \
805 "$base_url/ajax.js" \
806 "$base_url/table/table-sort.js" \
807 "$base_url/table/table-sort.css" \
808 "$base_url/base.css"\
809
810 table-sort-begin 'width60'
811
812 cat <<EOF
813 <p id="home-link">
814 <a href="/">oilshell.org</a>
815 </p>
816
817 <h1>Wedge Builds</h1>
818EOF
819
820 tsv2html3 $tasks_tsv
821
822 cat <<EOF
823 <p>
824 <a href="tasks.tsv">tasks.tsv</a>
825 </p>
826EOF
827
828 table-sort-end 'tasks' # ID for sorting
829}
830
831NPROC=$(nproc)
832#NPROC=1
833
834install-wedge-list() {
835 ### Reads task rows from stdin
836 local parallel=${1:-}
837
838 mkdir -p _build/wedge/logs
839
840 local -a flags
841 if test -n "$parallel"; then
842 log ""
843 log "=== Installing wedges with $NPROC jobs in parallel"
844 log ""
845 flags=( -P $NPROC )
846 else
847 log ""
848 log "=== Installing wedges serially"
849 log ""
850 fi
851
852 # Reads from stdin
853 # Note: --process-slot-var requires GNU xargs! busybox args doesn't have it.
854 #
855 # $name $version $wedge_dir
856 xargs "${flags[@]}" -n 3 --process-slot-var=XARGS_SLOT -- $0 maybe-install-wedge
857
858 #xargs "${flags[@]}" -n 3 --process-slot-var=XARGS_SLOT -- $0 dummy-task-wrapper
859}
860
861write-task-report() {
862 local tasks_tsv=_build/wedge/logs/tasks.tsv
863
864 python3 devtools/tsv_concat.py $WEDGE_LOG_DIR/*.task.tsv > $tasks_tsv
865 log "Wrote $tasks_tsv"
866
867 # TODO: version can be right-justified?
868 here-schema-tsv-4col >_build/wedge/logs/tasks.schema.tsv <<EOF
869column_name type precision strftime
870status integer 0 -
871elapsed_secs float 1 -
872user_secs float 1 -
873start_time float 1 %H:%M:%S
874end_time float 1 %H:%M:%S
875sys_secs float 1 -
876max_rss_KiB integer 0 -
877xargs_slot integer 0 -
878wedge string 0 -
879wedge_HREF string 0 -
880version string 0 -
881EOF
882
883 index-html $tasks_tsv > $WEDGE_LOG_DIR/index.html
884 log "Wrote $WEDGE_LOG_DIR/index.html"
885}
886
887install-spec-bin-fast() {
888 spec-bin-wedges | install-wedge-list T
889 write-task-report
890}
891
892fake-py3-libs-wedge() {
893 local name=py3-libs
894 local version=$PY3_LIBS_VERSION
895
896 local task_file=$WEDGE_LOG_DIR/$name.task.tsv
897 local log_file=$WEDGE_LOG_DIR/$name.log.txt
898
899 my-time-tsv --print-header \
900 --field xargs_slot \
901 --field wedge \
902 --field wedge_HREF \
903 --field version \
904 --output $task_file
905
906 # There is no xargs slot!
907 my-time-tsv \
908 --field "-1" \
909 --field "$name" \
910 --field "$name.log.txt" \
911 --field "$version" \
912 --append \
913 --output $task_file \
914 $0 install-py3-libs >$log_file 2>&1 || true
915
916 echo " FAKE $(timestamp) $name $version"
917}
918
919install-wedges-fast() {
920 local extra=${1:-}
921
922 # For contributor setup: we need to use this BEFORE running build/py.sh all
923 build/py.sh time-helper
924
925 echo " START $(timestamp)"
926
927 # Do all of them in parallel
928 if test -n "$extra"; then
929 { contributor-wedges; extra-wedges; } | install-wedge-list T
930 else
931 contributor-wedges | install-wedge-list T
932 fi
933
934 fake-py3-libs-wedge
935 echo " END $(timestamp)"
936
937 write-task-report
938}
939
940install-wedges-soil() {
941 install-wedges-fast extra
942}
943
944#
945# Unboxed wedge builds
946#
947
948uftrace-host() {
949 ### built on demand; run $0 first
950
951 # BUG: doesn't detect python3
952 # WEDGE tells me that it depends on pkg-config
953 # 'apt-get install pkgconf' gets it
954 # TODO: Should use python3 WEDGE instead of SYSTEM python3?
955 deps/wedge.sh unboxed _build/deps-source/uftrace
956}
957
958bloaty-host() {
959 deps/wedge.sh unboxed _build/deps-source/bloaty
960}
961
962R-libs-host() {
963 deps/wedge.sh unboxed _build/deps-source/R-libs
964}
965
966#
967# Wedges built inside a container, for copying into a container
968#
969
970boxed-wedges() {
971 #### host _build/wedge/binary -> guest container /wedge or ~/wedge
972
973 #export-podman
974
975 # TODO:
976 #
977 # - Add equivalents of spec-bin
978 # - Use the same manifest as install-wedges-fast
979 # - so then you can delete the _build/wedge dir to re-run it
980 # - use xargs -n 1 so it's done serially
981
982 # - Do these lazily like we do in install-wedges-fast?
983
984 # We can test if the dir _build/wedge/binary/oils-for-unix.org/pkg/FOO exists
985 # if wedge-exists "$name" "$version" "$wedge_dir"; then
986 # echo "CACHED $(timestamp) $name $version"
987 # return
988 # fi
989
990 if false; then
991 deps/wedge.sh boxed deps/source.medo/time-helper
992 deps/wedge.sh boxed deps/source.medo/cmark/
993 deps/wedge.sh boxed deps/source.medo/re2c/
994 deps/wedge.sh boxed deps/source.medo/python3/
995 fi
996
997 if false; then
998 deps/wedge.sh boxed deps/source.medo/bloaty/
999 fi
1000
1001 if true; then
1002 # build with debian-12, because soil-benchmarks2 is, because it has R
1003 deps/wedge.sh boxed deps/source.medo/uftrace/ '' debian-12
1004 # python2 needed everywhere
1005 #deps/wedge.sh boxed deps/source.medo/python2/ '' debian-12
1006
1007 # TODO: build with debian-12
1008 # Used in {benchmarks,benchmarks2,other-tests}
1009 #deps/wedge.sh boxed deps/source.medo/R-libs/ '' debian-12
1010 fi
1011}
1012
1013boxed-spec-bin() {
1014 if true; then
1015 #deps/wedge.sh boxed deps/source.medo/bash '4.4'
1016 deps/wedge.sh boxed deps/source.medo/bash '5.2.21'
1017 fi
1018
1019 if false; then
1020 deps/wedge.sh boxed deps/source.medo/dash
1021 deps/wedge.sh boxed deps/source.medo/mksh
1022 fi
1023
1024 if false; then
1025 # Note: zsh requires libncursesw5-dev
1026 #deps/wedge.sh boxed deps/source.medo/zsh
1027
1028 #deps/wedge.sh boxed deps/source.medo/busybox
1029
1030 # Problem with out of tree build, as above. Skipping for now
1031 deps/wedge.sh boxed deps/source.medo/yash
1032 echo
1033 fi
1034}
1035
1036#
1037# Report
1038#
1039
1040commas() {
1041 # Wow I didn't know this :a trick
1042 #
1043 # OK this is a label and a loop, which makes sense. You can't do it with
1044 # pure regex.
1045 #
1046 # https://shallowsky.com/blog/linux/cmdline/sed-improve-comma-insertion.html
1047 # https://shallowsky.com/blog/linux/cmdline/sed-improve-comma-insertion.html
1048 sed ':a;s/\b\([0-9]\+\)\([0-9]\{3\}\)\b/\1,\2/;ta'
1049}
1050
1051wedge-sizes() {
1052 local tmp=_tmp/wedge-sizes.txt
1053
1054 # -b is --bytes, but use short flag for busybox compat
1055 du -s -b /wedge/*/*/* ~/wedge/*/*/* | awk '
1056 { print $0 # print the line
1057 total_bytes += $1 # accumulate
1058 }
1059END { print total_bytes " TOTAL" }
1060' > $tmp
1061
1062 # printf justifies du output
1063 cat $tmp | commas | xargs -n 2 printf '%15s %s\n'
1064 echo
1065
1066 #du -s --si /wedge/*/*/* ~/wedge/*/*/*
1067 #echo
1068}
1069
1070wedge-report() {
1071 # 4 levels deep shows the package
1072 if command -v tree > /dev/null; then
1073 tree -L 4 /wedge ~/wedge
1074 echo
1075 fi
1076
1077 wedge-sizes
1078
1079 local tmp=_tmp/wedge-manifest.txt
1080
1081 echo 'Biggest files'
1082 if ! find /wedge ~/wedge -type f -a -printf '%10s %P\n' > $tmp; then
1083 # busybox find doesn't have -printf
1084 echo 'find -printf failed'
1085 return
1086 fi
1087
1088 set +o errexit # ignore SIGPIPE
1089 sort -n --reverse $tmp | head -n 20 | commas
1090 set -o errexit
1091
1092 echo
1093
1094 # Show the most common file extensions
1095 #
1096 # I feel like we should be able to get rid of .a files? That's 92 MB, second
1097 # most common
1098 #
1099 # There are also duplicate .a files for Python -- should look at how distros
1100 # get rid of those
1101
1102 cat $tmp | python3 -c '
1103import os, sys, collections
1104
1105bytes = collections.Counter()
1106files = collections.Counter()
1107
1108for line in sys.stdin:
1109 size, path = line.split(None, 1)
1110 path = path.strip() # remove newline
1111 _, ext = os.path.splitext(path)
1112 size = int(size)
1113
1114 bytes[ext] += size
1115 files[ext] += 1
1116
1117#print(bytes)
1118#print(files)
1119
1120n = 20
1121
1122print("Most common file types")
1123for ext, count in files.most_common()[:n]:
1124 print("%10d %s" % (count, ext))
1125
1126print()
1127
1128print("Total bytes by file type")
1129for ext, total_bytes in bytes.most_common()[:n]:
1130 print("%10d %s" % (total_bytes, ext))
1131' | commas
1132}
1133
1134task-five "$@"