source | all docs for version 0.7.0 | all versions | oilshell.org
Warning: Work in progress! Leave feedback on Zulip or Github if you'd like this doc to be updated.
Parse Error: Can be determined statically spec/parse-errors.test.sh
TODO: See test/runtime-errors.sh. Merge them here.
Fatal Error: terminates the interpreter unconditionally, e.g. divide by zero does this in bash.
Non-fatal error: terminates the current builtin and exits 1
non-fatal errors can be turned into fatal errors.
by Strict modes: set -o errexit
strict modes can also things that are not errors at all into fatal errors set -o nounset set -o failglob
Fatal errors can be turned into non-fatal ones!!!!
by dparen:
(( 1 / 0 ))
by command sub -- although this involves another process so it's understandable!
set -o errexit echo $(exit 1)
strict_array strict_errexit strict_arith
TODO: strict-word-eval? for unicode errors for subshell negative indices? I think this is most consistent right now.
TODO:
p_die() internally
w = w_parser.ReadWord()
if w is None:
do something with w_parser.Error()
Related to memory management API:
# arena is the out param
arena = pool.NewArena()
c_parser = cmd_parse.CommandParser(w_parser, arena)
bool ok = c_parser.Parse()
if ok:
arena.RootNode() # turns indexes into pointers?
arena.Deallocate() # d
else:
c_parser.Error() # Is this still a stack?
Idea:
assign each one of these a code, and decide what to do based on a table?
then have an error CONTEXT
based on spec tests?
and error context takes an error code, looks it up in a table, and decides whether to catch or to reraise!
List of contexts:
echo $(( 1 / 0 )) echo 'after-$(()) (( 1 / 0 )) echo 'after-$(())
Divide by zero: $(( 1 / 0 ))
^
Maybe: integer overflow. But we want big numbers.
Type errors between integers and strings:
x=foo
$(( x * 2 )) # doesn't make sense, except in bash's crazy world.
Invalid hex constant:
x=0xabcg
echo $(( x * 2 )) (fatal in bash)
regcomp parse error:
x=$(cat invalid-syntax.txt) [[ foo =~ $x ]]
IMPORTANT: Command sub error $(exit 1)
User-requested error: ${undef?error}
set -o nounset
def _EmptyStrOrError(self, val, token=None):
# calls `e_die()`
Variants: nounset: index out of bounds ${a[3]} I guess same diagnostic?
In bash you can set an index out of bounds, like
b[2]=9
Might want to have a mode for this?
set -o failglob TODO: not implemented might need PWD diagnostic
Redirects: Redirect to empty filename/descriptor ( or array)
{ break; }
^~~~~~ break only invalid inside loop, etc.
NotImplementedError
e.g for var ref ${!a}
bash associative arrays? I think we want most of that
$"" ?
|& not yet done
;;& for case -- although parsing it is all of the work I guess
some could be parse time errors too though?
String Slicing and String Length require valid utf-8 characters
s=$(cat invalid.txt) echo ${#s} # code points echo ${s:1:3} # code opints
Slicing: Index is negative. ${foo: -4} and ${foo: 1 : -4} aren't supported right now, unlike bash and zsh.
IMPORTANT: subshell error ( exit 1 )
set -o errexit -- turns NON-FATAL error into FATAL error.
set -o pipefail pipefail might need some fanciness for ${PIPESTATUS}
Trying to set readonly variable: readonly foo=bar foo=x (could any of this be done at compile time?)
Trying to redeclare a variable? That can also be parse time. local x=1 local x=2
Type errors between Str and StrArray: -- strict-array controls this EvalWordToString calls e_die()`
echo foo > "$@" ^-- # Should have what it evaluated to? # This could be static too
case "$@" in "$@") echo bad;; esac
${undef:-"$@"} is OK, but ${var%"$@"} doesn't make sense really. ${v/"$@"/"$@"}
LHS evaluation: s='abc' s[1]=X # invalid because it's a string, not an array
Invalid descriptor:
fd=$(cat invalid.txt) echo foo 2>& $fd
In core/builtins.py:
util.usage('...')
return 1
A usage error is a runtime error that results in the builtin returning 1.
Builtin has too many arguments -- but this falls under the errexit rule cd foo bar baz continue "$@" (Parse error: continue 1 2 3)
Although we might want to highlight the extra args.
Fatal error from system calls: fork() could fail in theory
Some are not failures:
stat() [[ -f /tmp/foo ]]
cd /ff chdir() # exit code 1
cat <nonexistent # This is just exit code 1
Runtime: Stack Too Deep (catch infinite recursion) Out of memory: should not happen with OSH, but maybe with Oil
The way bash works 0x$var can be a hex literal. so var=xx makes this invalid. hex/octal/decimal have this problem.
regcomp() errors (sometimes at parse time; other times at runtime)
Need to show stack trace for "source" like Python. Prototype this.
Also might show which token thing caused you to be in arith parse state, like:
$((echo hi)) ^~ ^~ Arith Invalid token