View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001046 | 1003.1(2013)/Issue7+TC1 | Shell and Utilities | public | 2016-04-23 16:09 | 2017-06-01 15:09 |
| Reporter | rillig | Assigned To | |||
| Priority | normal | Severity | Objection | Type | Error |
| Status | Closed | Resolution | Withdrawn | ||
| Name | Roland Illig | ||||
| Organization | |||||
| User Reference | |||||
| Section | 2.10. Shell Grammar | ||||
| Page Number | unsure | ||||
| Line Number | unsure | ||||
| Interp Status | --- | ||||
| Final Accepted Text | |||||
| Summary | 0001046: Grammar production if_clause allows "if command then" | ||||
| Description | http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_10_02 According to that grammar, the following text is valid: if echo then echo yes else echo no fi This is because the ''if_clause'' refers to the ''compound_list'', which, in its simplest form, can be reduced to just a WORD token, using these productions: if_clause := "if" compound_list "then" compound_list "fi" compound_list := term term := and_or and_or := pipeline pipeline := pipe_sequence pipe_sequence := simple_command simple_command := cmd_name cmd_name := WORD All shells that I know interpret this differently, taking the ''then echo yes else echo no fi'' as arguments to the ''echo'' command. | ||||
| Desired Action | Check whether the grammar in its current form really allows this interpretation. If it does, restrict the grammar to disallow this, since it creates ambiguous programs. | ||||
| Tags | No tags attached. | ||||
| related to | 0001145 | Closed | 1003.1(2016/18)/Issue7+TC2 | compound-list does not required a terminating semicolon even in compound commands, which is contradictory with implementations. |
|
|
In if echo then echo yes else echo no fi once "echo" (the fist one) has been parsed as a command name, none of the following args is elegible to be a reserved word, until the command is finished (which takes some other syntax character, ';' or \n or perhaps ) in appropriate circumstances. Hence the interpretation you're suggesting cannot occur as none of "then" "else" or "fi" in that example are the tokens with those names, they are all just words. This isn't because of the grammar as such, but because of the somewhat strange way the lexical analyser is required to work. |
|
|
No, the grammar does not permit that by Rule 1 in that section. Keyword recognition, due to the if opening the context, has priority over assignment as a utility argument WORD, and has an implied separator_op of ';' preceding for the Then, Elif, and Else keyword TOKENs as a result terminating the cmd_suffix of the echo term. Shells treating the then... clause as arguments are buggy therefore. |
|
|
Rule 1 applies only to command names, and there is no such thing as an implied separator_op. Perhaps the text needs to be made clearer about this, but I am quite certain that reserved words (with "in" as something of a special case) are only ever intended to be detected as reserved words at the beginning of a command (after some kind of command ending delimiter). |
|
|
Yes, it could be stated better, but at the point cmd_suffix might start the Then as a continuation of the if_clause has priority in determining that the cmd_name alone is what satisfies simple_command being complete. |
|
|
Thanks, kre, for the explanation. It is a tricky topic, but now that I have thought about it for several hours, the specification contains all the necessary details. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2016-04-23 16:09 | rillig | New Issue | |
| 2016-04-23 16:09 | rillig | Name | => Roland Illig |
| 2016-04-23 16:09 | rillig | Section | => 2.10. Shell Grammar |
| 2016-04-23 16:09 | rillig | Page Number | => unsure |
| 2016-04-23 16:09 | rillig | Line Number | => unsure |
| 2016-04-23 17:02 | kre | Note Added: 0003168 | |
| 2016-04-23 17:02 | shware_systems | Note Added: 0003169 | |
| 2016-04-23 17:17 | shware_systems | Note Edited: 0003169 | |
| 2016-04-23 17:38 | shware_systems | Note Edited: 0003169 | |
| 2016-04-23 17:50 | kre | Note Added: 0003170 | |
| 2016-04-23 18:27 | shware_systems | Note Added: 0003171 | |
| 2016-04-23 19:59 | rillig | Note Added: 0003172 | |
| 2016-04-25 08:55 | geoffclare | Project | 2008-TC2 => 1003.1(2013)/Issue7+TC1 |
| 2017-06-01 15:09 | geoffclare | Interp Status | => --- |
| 2017-06-01 15:09 | geoffclare | Status | New => Closed |
| 2017-06-01 15:09 | geoffclare | Resolution | Open => Withdrawn |
| 2017-06-15 08:55 | geoffclare | Relationship added | related to 0001145 |
