View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001425 | 1003.1(2016/18)/Issue7+TC2 | Shell and Utilities | public | 2020-11-23 23:10 | 2024-06-11 09:08 |
| Reporter | kre | Assigned To | |||
| Priority | normal | Severity | Objection | Type | Error |
| Status | Closed | Resolution | Accepted As Marked | ||
| Name | Robert Elz | ||||
| Organization | |||||
| User Reference | |||||
| Section | XCU 2.14 / exit | ||||
| Page Number | 2399 | ||||
| Line Number | 76764-5 | ||||
| Interp Status | --- | ||||
| Final Accepted Text | 0001425:0005266 | ||||
| Summary | 0001425: exit in an EXIT trap is not clearly specified | ||||
| Description | the page for the special builtin command "exit" says A trap on EXIT shall be executed before the shell terminates, except when the exit utility is invoked in that trap itself, in which case the shell shall exit immediately. The issue is what "that trap itself" means. Consider $SHELL -c 'trap "trap \"echo bar\" EXIT; echo foo; exit" EXIT' When the exit command in the outer exit trap is executed, the EXIT trap has been changed to be something different, that is, the shell is not executing the (current) EXIT trap when that exit command is executed, so by a literal reading of the standard, the inner trap should fire, and the shell should print foo bar before exiting. That is what the FreeBSD and NetBSD shells, and bosh, mksh, and zsh do. bash yash ksh93 and dash print only "foo", that is they seem to ignore the "that" in "that exit trap" and treat the text as if it said except when the exit utility is invoked in an exit trap The page in XCU 2.14 for trap says nothing pertinent to this issue (I am not suggesting that it should, just noting it). | ||||
| Desired Action | Since the "reference" shells seem to do this the broken way, it is unlikely that the standard will have words (or an example) added to reinforce the current wording (I am guessing). But since the other shells are doing exactly what the standard requires of them, it would be incorrect, and unfair to specify this so as to require the currently non-standard behaviour that the reference shells implement. Given that, the best that can be done, I believe, is to add a new sentence, immediately after the one quoted above (which is unchanged in 202x drafts) something like It is unspecified whether an EXIT trap defined while executing a previous EXIT trap will be executed before the shell terminates after completing execution (via an exit command, or reaching the end) of the earlier EXIT trao. | ||||
| Tags | tc3-2008 | ||||
|
|
In c181 on page 2399 append another sentence to the paragraph at lines 76764-76765:It is unspecified whether setting a new trap action on EXIT during execution of a trap action on EXIT will cause the new trap action to be executed before the shell terminates. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2020-11-23 23:10 | kre | New Issue | |
| 2020-11-23 23:10 | kre | Name | => Robert Elz |
| 2020-11-23 23:10 | kre | Section | => XCU 2.14 / exit |
| 2020-11-23 23:10 | kre | Page Number | => 2399 |
| 2020-11-23 23:10 | kre | Line Number | => 76764-5 |
| 2021-03-11 17:07 | rhansen | Note Added: 0005266 | |
| 2021-03-11 17:08 | rhansen | Interp Status | => --- |
| 2021-03-11 17:08 | rhansen | Final Accepted Text | => 0001425:0005266 |
| 2021-03-11 17:08 | rhansen | Status | New => Resolved |
| 2021-03-11 17:08 | rhansen | Resolution | Open => Accepted As Marked |
| 2021-03-11 17:09 | rhansen | Tag Attached: tc3-2008 | |
| 2021-03-25 14:27 | geoffclare | Status | Resolved => Applied |
| 2024-06-11 09:08 | agadmin | Status | Applied => Closed |
