View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001069 | 1003.1(2013)/Issue7+TC1 | System Interfaces | public | 2016-08-23 08:22 | 2018-01-04 17:01 |
| Reporter | dannyniu | Assigned To | |||
| Priority | normal | Severity | Objection | Type | Error |
| Status | Closed | Resolution | Rejected | ||
| Name | DannyNiu/NJF | ||||
| Organization | |||||
| User Reference | |||||
| Section | System Interfaces, waitid | ||||
| Page Number | 2212 | ||||
| Line Number | 70253 | ||||
| Interp Status | --- | ||||
| Final Accepted Text | Rejected for the reasons noted in 0001069:0003906. | ||||
| Summary | 0001069: with a process group ID equal to (/*are you sure?*/pid_t)id | ||||
| Description | In the description for waitid, it says if idtype is P_PGID, then the second argument is casted to type pid_t, which is the integer type for process IDs, but idtype says it's group ID. I'm confused. | ||||
| Desired Action | Change the line to: If idtype is P_PGID, waitid( ) shall wait for any child with a process group ID equal to (gid_t)id. | ||||
| Tags | No tags attached. | ||||
|
|
No, pid_t is correct. We're talking of process groups here (as in setpgid()), not of user group owner (egid, gid as set with setegid()/setgid()). We're waiting for processes in a group of processes, not processes owned by some user group. |
|
|
Re 3355. Okay, my mindlessness. I took a look at FreeBSD man page in the mean time, would it worth preventing future re-reporting by mentioning the reason it's id_t not pid_t is because implementations may provide addition P_*ID that are not necessarily PIDs? |
|
|
If you just look at <sys/types.h> on P401, L13461 where pid_t's uses are listed, it says: "Used for process IDs and process group IDs." A group and a process group are two different things. A group ID and a process group ID are two different things. A gid_t and a pid_t are two different things. A group and a group ID and a gid_t are related. A process group and a process group ID and a pid_t are related. |
|
|
So, what if we quote the words "process group"? So that readers unfamiliar with the concepts of various IDs within the spec won't confuse it with real/effective/etc. group IDs of the process? Because honestly, I feel the words "process group ID" do need a bit of disambiguation to some people. |
|
|
Re: 0001069:0003358: I am totally against quoting this reference to "process group ID" unless we also quote all of the other uses of the other 447 defined terms in XBD section 3 of the standard. In just plain English text, it should be clear to anyone reading the text: "If idtype is P_PGID, waited() shall wait for any child with a process group ID equal to (pid_t)id." that "process group ID" is referring to the defined term "process group ID" and is not the referring to the defined term "group ID". It is referring to the ID of a group of processes ("process group ID"). There is no possessive that would indicate that there is a reference to a "group ID" of a "process" which would be referenced as a "process"'s "real group ID" or a "process"'s "effective group ID". |
|
|
Re: 3359 I'm not in favor of quoting them either. Some online documents hyperlink each occurrence of words or phrases in a Glossary or Definitions section to the definition, though, which I think could be useful. The HTML version does this for Margin Codes, pops up a window with the definitions. Doing this for all of XBD3, at this stage, would be more effort than quoting would, however. If a change were to be made I'd change process group to process-group, so people don't assume a possessive <'s> was omitted accidentally. It's not that it should be clear, but that some readers will do this anyways because they don't bother to read XBD 3 or <types.h> to note pid_t applies to individual processes and process groups, or forget it is one the 300+ entries in that section. |
|
|
We do not want to make any special markings just on this page for one or two defined terms. We have, however, suggested to the editor that it would be nice if all uses of terms defined in the standard could be presented as links that, if clicked, would take you to the definition of that term in the XBD volume of the standard. Since this is a huge amount of work for the editor, we do not know if this suggestion will be implemented any time soon, or at all. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2016-08-23 08:22 | dannyniu | New Issue | |
| 2016-08-23 08:22 | dannyniu | Name | => DannyNiu/NJF |
| 2016-08-23 08:22 | dannyniu | Section | => System Interfaces, waitid |
| 2016-08-23 08:22 | dannyniu | Page Number | => 2212 |
| 2016-08-23 08:22 | dannyniu | Line Number | => 70253 |
| 2016-08-23 08:50 | stephane | Note Added: 0003355 | |
| 2016-08-23 11:05 | dannyniu | Note Added: 0003356 | |
| 2016-08-23 19:48 | Don Cragun | Note Added: 0003357 | |
| 2016-08-24 02:01 | dannyniu | Note Added: 0003358 | |
| 2016-08-24 02:05 | dannyniu | Note Edited: 0003358 | |
| 2016-08-24 07:16 | Don Cragun | Note Added: 0003359 | |
| 2017-12-16 20:32 | shware_systems | Note Added: 0003903 | |
| 2018-01-04 17:00 | Don Cragun | Note Added: 0003906 | |
| 2018-01-04 17:01 | Don Cragun | Interp Status | => --- |
| 2018-01-04 17:01 | Don Cragun | Final Accepted Text | => Rejected for the reasons noted in 0001069:0003906. |
| 2018-01-04 17:01 | Don Cragun | Status | New => Closed |
| 2018-01-04 17:01 | Don Cragun | Resolution | Open => Rejected |
