---
- branch: MAIN
  date: Sun Aug 19 23:50:27 UTC 2018
  files:
  - new: '1.21'
    old: '1.20'
    path: src/bin/sh/error.h
    pathrev: src/bin/sh/error.h@1.21
    type: modified
  - new: '1.21'
    old: '1.20'
    path: src/bin/sh/input.h
    pathrev: src/bin/sh/input.h@1.21
    type: modified
  - new: '1.159'
    old: '1.158'
    path: src/bin/sh/eval.c
    pathrev: src/bin/sh/eval.c@1.159
    type: modified
  - new: '1.63'
    old: '1.62'
    path: src/bin/sh/input.c
    pathrev: src/bin/sh/input.c@1.63
    type: modified
  - new: '1.75'
    old: '1.74'
    path: src/bin/sh/main.c
    pathrev: src/bin/sh/main.c@1.75
    type: modified
  - new: '1.150'
    old: '1.149'
    path: src/bin/sh/parser.c
    pathrev: src/bin/sh/parser.c@1.150
    type: modified
  - new: '1.45'
    old: '1.44'
    path: src/bin/sh/trap.c
    pathrev: src/bin/sh/trap.c@1.45
    type: modified
  - new: '1.23'
    old: '1.22'
    path: src/bin/sh/trap.h
    pathrev: src/bin/sh/trap.h@1.23
    type: modified
  id: 20180819T235027Z.52abba14560b85336a571b2be65534d49e25530b
  log: "PR bin/48875 (is related, and ameliorated, but not exactly \"fixed\")\n\nImport
    a whole set of tree evaluation enhancements from FreeBSD.\n\nWith these, before
    forking, the shell predicts (often) when all it will\nhave to do after forking
    (in the parent) is wait for the child and then\nexit with the status from the
    child, and in such a case simply does not\nfork, but rather allows the child to
    take over the parent's role.\n\nThis turns out to handle the particular test case
    from PR bin/48875 in\nsuch a way that it works as hoped, rather than as it did
    (the delay there\nwas caused by an extra copy of the shell hanging around waiting
    for the\nbackground child to complete ... and keeping the command substitution\nstdout
    open, so the \"real\" parent had to wait in case more output appeared).\n\nAs
    part of doing this, redirection processing for compound commands gets\nmoved out
    of evalsubshell() and into a new evalredir(), which allows us\nto properly handle
    errors occurring while performing those redirects,\nand not mishandle (as in simply
    forget) fd's which had been moved out\nof the way temporarily.\n\nevaltree() has
    its degree of recursion reduced by making it loop to\nhandle the subsequent operation:
    that is instead of (for any binop\nlike ';' '&&' (etc)) where it used to\n\tevaltree(node->left);\n\tevaltree(node->right);\n\treturn;\nit
    now does (kind of)\n\tnext = node;\n\twhile ((node = next) != NULL) {\n\t\tnext
    = NULL;\n\n\t\tif (node is a binary op) {\n\t\t\tevaltree(node->left);\n\t\t\tif
    appropriate /* if && test for success, etc */\n\t\t\t\tnext = node->right;\n\t\t\tcontinue;\n\t\t}\n\t\t/*
    similar for loops, etc */\n\t}\nwhich can be a good saving, as while the left
    side (now) tends to be\n(usually) a simple (or simpleish) command, the right side
    can be many\ncommands (in a command sequence like a; b; c; d; ...  the node at
    the\ntop of the tree will now have \"a\" as its left node, and the tree for\nb;
    c; d; ... as its right node - until now everything was evaluated\nrecursively
    so it made no difference, and the tree was constructed\nthe other way).\n\nif/while/...
    statements are done similarly, recurse to evaluate the\ncondition, then if the
    (or one of the) body parts is to be evaluated,\nset next to that, and loop (previously
    it recursed).\n\nThere is more to do in this area (particularly in the way that
    case\nstatements are processed - we can avoid recursion there as well) but\nthat
    can wait for another day.\n\nWhile doing all of this we keep much better track
    of when the shell is\njust going to exit once the current tree is evaluated (with
    a new\npredicate at_eof() to tell us that we have, for sure, reached the end\nof
    the input stream, that is, this shell will, for certain, not be reading\nmore
    command input) and use that info to avoid unneeded forks.   For that\nwe also
    need another new predicate (have_traps()) to determine of there\nare any caught
    traps which might occur - if there are, we need to remain\nto (potentially) handle
    them, so these optimisations will not occur (to\nmake the issue in PR 48875 appear
    again, run the same code, but with a\ntrap set to execute some code when a signal
    (or EXIT) occurs - note that\nthe trap must be set in the appropriate level of
    sub-shell to have this\neffect, any caught traps are cleared in a subshell whenever
    one is created).\n\nThere is still work to be done to handle traps properly, whatever\nweirdness
    they do (some of which is related to some of this.)\n\nThese changes do not need
    man page updates, but 48875 does - an update\nto sh.1 will be forthcoming once
    it is decided what it should say...\n\nOnce again, all the heavy lifting for this
    set of changes comes directly\n(with thanks) from the FreeBSD shell.\n\nXXX pullup-8
    (but not very soon)\n"
  module: src
  subject: 'CVS commit: src/bin/sh'
  unixtime: '1534722627'
  user: kre