aboutsummaryrefslogtreecommitdiffstats
path: root/erts/emulator/beam/ops.tab
diff options
context:
space:
mode:
authorBjörn Gustavsson <[email protected]>2017-11-14 09:05:31 +0100
committerBjörn Gustavsson <[email protected]>2017-11-30 12:46:51 +0100
commitef2e55c707a6c3d1f54f2acb9c8bfbe3fb7a8659 (patch)
tree7a9eae08bd786efbebddb53763888a66ed580abb /erts/emulator/beam/ops.tab
parent0c4574cc31414a5204eabfad7eb0ac5095c0ace1 (diff)
downloadotp-ef2e55c707a6c3d1f54f2acb9c8bfbe3fb7a8659.tar.gz
otp-ef2e55c707a6c3d1f54f2acb9c8bfbe3fb7a8659.tar.bz2
otp-ef2e55c707a6c3d1f54f2acb9c8bfbe3fb7a8659.zip
Add syntax in try/catch to retrieve the stacktrace directly
This commit adds a new syntax for retrieving the stacktrace without calling erlang:get_stacktrace/0. That allow us to deprecate erlang:get_stacktrace/0 and ultimately remove it. The problem with erlang:get_stacktrace/0 is that it can keep huge terms in a process for an indefinite time after an exception. The stacktrace can be huge after a 'function_clause' exception or a failed call to a BIF or operator, because the arguments for the call will be included in the stacktrace. For example: 1> catch abs(lists:seq(1, 1000)). {'EXIT',{badarg,[{erlang,abs, [[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20|...]], []}, {erl_eval,do_apply,6,[{file,"erl_eval.erl"},{line,674}]}, {erl_eval,expr,5,[{file,"erl_eval.erl"},{line,431}]}, {shell,exprs,7,[{file,"shell.erl"},{line,687}]}, {shell,eval_exprs,7,[{file,"shell.erl"},{line,642}]}, {shell,eval_loop,3,[{file,"shell.erl"},{line,627}]}]}} 2> erlang:get_stacktrace(). [{erlang,abs, [[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22, 23,24|...]], []}, {erl_eval,do_apply,6,[{file,"erl_eval.erl"},{line,674}]}, {erl_eval,expr,5,[{file,"erl_eval.erl"},{line,431}]}, {shell,exprs,7,[{file,"shell.erl"},{line,687}]}, {shell,eval_exprs,7,[{file,"shell.erl"},{line,642}]}, {shell,eval_loop,3,[{file,"shell.erl"},{line,627}]}] 3> We can extend the syntax for clauses in try/catch to optionally bind the stacktrace to a variable. Here is an example using the current syntax: try Expr catch C:E -> Stk = erlang:get_stacktrace(), . . . In the new syntax, it would look like: try Expr catch C:E:Stk -> . . . Only a variable (not a pattern) is allowed in the stacktrace position, to discourage matching of the stacktrace. (Matching would also be expensive, because the raw format of the stacktrace would have to be converted to the cooked form before matching.) Note that: try Expr catch E -> . . . is a shorthand for: try Expr catch throw:E -> . . . If the stacktrace is to be retrieved for a throw, the 'throw:' prefix must be explicitly included: try Expr catch throw:E:Stk -> . . .
Diffstat (limited to 'erts/emulator/beam/ops.tab')
-rw-r--r--erts/emulator/beam/ops.tab6
1 files changed, 6 insertions, 0 deletions
diff --git a/erts/emulator/beam/ops.tab b/erts/emulator/beam/ops.tab
index 3df91056cb..ba4ca67e5e 100644
--- a/erts/emulator/beam/ops.tab
+++ b/erts/emulator/beam/ops.tab
@@ -1575,3 +1575,9 @@ i_recv_mark
recv_set Fail | label Lbl | loop_rec Lf Reg => \
i_recv_set | label Lbl | loop_rec Lf Reg
i_recv_set
+
+#
+# OTP 21.
+#
+
+build_stacktrace