Re: zsh: Strange feature with 'jobs' commmand

Re: zsh: Strange feature with 'jobs' commmand

[Peter Stephenson]

=?iso-8859-1?Q?Bj=F6rn_Johannesson?= wrote:
> Hi.
> 
> If you start a few jobs in the background and do:
> zsh% jobs
> [1]  - running    xmixer
> [2]  + running    gtcd
> zsh% _
> 
> ok, this is expected but this is not...
> zsh% jobs | less
> (END)

This annoys everybody.

The problem is fairly localised: you only need to list jobs in
subshells, and the subshells will not be changing the status of jobs.
So maybe something relatively straightforward like the following is good
enough.

Index: Src/exec.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/exec.c,v
retrieving revision 1.41
diff -u -r1.41 exec.c
--- Src/exec.c	6 May 2002 14:46:11 -0000	1.41
+++ Src/exec.c	30 May 2002 15:12:37 -0000
@@ -2507,13 +2507,13 @@
 static void
 entersubsh(int how, int cl, int fake)
 {
-    int sig;
+    int sig, monitor;
 
     if (cl != 2)
 	for (sig = 0; sig < VSIGCOUNT; sig++)
 	    if (!(sigtrapped[sig] & ZSIG_FUNC))
 		unsettrap(sig);
-    if (unset(MONITOR)) {
+    if (!(monitor = isset(MONITOR))) {
 	if (how & Z_ASYNC) {
 	    settrap(SIGINT, NULL);
 	    settrap(SIGQUIT, NULL);
@@ -2569,7 +2569,7 @@
     opts[MONITOR] = opts[USEZLE] = 0;
     zleactive = 0;
     if (cl)
-	clearjobtab();
+	clearjobtab(monitor);
     times(&shtms);
     forklevel = locallevel;
 }
Index: Src/jobs.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/jobs.c,v
retrieving revision 1.15
diff -u -r1.15 jobs.c
--- Src/jobs.c	21 Dec 2001 16:33:39 -0000	1.15
+++ Src/jobs.c	30 May 2002 15:12:37 -0000
@@ -55,6 +55,12 @@
 /**/
 mod_export struct job jobtab[MAXJOB];
 
+/* If we have entered a subshell, the original shell's job table. */
+static struct job *oldjobtab;
+
+/* The size of that. */
+static int oldmaxjob;
+
 /* shell timings */
  
 /**/
@@ -612,13 +618,18 @@
 printjob(Job jn, int lng, int synch)
 {
     Process pn;
-    int job = jn - jobtab, len = 9, sig, sflag = 0, llen;
+    int job, len = 9, sig, sflag = 0, llen;
     int conted = 0, lineleng = columns, skip = 0, doputnl = 0;
     FILE *fout = (synch == 2) ? stdout : shout;
 
     if (jn->stat & STAT_NOPRINT)
 	return;
 
+    if (jn < jobtab || jn >= jobtab + MAXJOB)
+	job = jn - oldjobtab;
+    else
+	job = jn - jobtab;
+
     if (lng < 0) {
 	conted = 1;
 	lng = 0;
@@ -655,10 +666,10 @@
 	}
     }
 
-/* print if necessary */
+/* print if necessary: ignore option state on explicit call to `jobs'. */
 
-    if (interact && jobbing && ((jn->stat & STAT_STOPPED) || sflag ||
-				job != thisjob)) {
+    if (((interact && jobbing) || synch == 2) &&
+	((jn->stat & STAT_STOPPED) || sflag ||job != thisjob)) {
 	int len2, fline = 1;
 	Process qn;
 
@@ -681,7 +692,7 @@
 	    if (job != thisjob) {
 		if (fline)
 		    fprintf(fout, "[%ld]  %c ",
-			    (long)(jn - jobtab),
+			    (long)job,
 			    (job == curjob) ? '+'
 			    : (job == prevjob) ? '-' : ' ');
 		else
@@ -956,13 +967,32 @@
 
 /**/
 mod_export void
-clearjobtab(void)
+clearjobtab(int monitor)
 {
     int i;
 
-    for (i = 1; i < MAXJOB; i++)
-	if (jobtab[i].ty)
+    for (i = 1; i < MAXJOB; i++) {
+	if (jobtab[i].ty) {
 	    zfree(jobtab[i].ty, sizeof(struct ttyinfo));
+	    jobtab[i].ty = NULL;
+	}
+	if (monitor) {
+	    /*
+	     * See if there is a jobtable worth saving.
+	     * We never free the saved version; it only happens
+	     * once for each subshell of a shell with job control,
+	     * so doesn't create a leak.
+	     */
+	    if (jobtab[i].stat)
+		oldmaxjob = i;
+	}
+    }
+
+    if (monitor && oldmaxjob) {
+	int sz = oldmaxjob * sizeof(struct job);
+	oldjobtab = (struct job *)zalloc(sz);
+	memcpy(oldjobtab, jobtab, sz);
+    }
 
     memset(jobtab, 0, sizeof(jobtab)); /* zero out table */
 }
@@ -1274,13 +1304,22 @@
 	    firstjob = curjob;
 	} else if (func == BIN_JOBS) {
 	    /* List jobs. */
-	    for (job = 0; job != MAXJOB; job++)
-		if (job != thisjob && jobtab[job].stat) {
+	    struct job *jobptr;
+	    int maxjob;
+	    if (unset(MONITOR) && oldmaxjob) {
+		jobptr = oldjobtab;
+		maxjob = oldmaxjob;
+	    } else {
+		jobptr = jobtab;
+		maxjob = MAXJOB;
+	    }
+	    for (job = 0; job != maxjob; job++, jobptr++)
+		if (job != thisjob && jobptr->stat) {
 		    if ((!ops['r'] && !ops['s']) ||
 			(ops['r'] && ops['s']) ||
-			(ops['r'] && !(jobtab[job].stat & STAT_STOPPED)) ||
-			(ops['s'] && jobtab[job].stat & STAT_STOPPED))
-			printjob(job + jobtab, lng, 2);
+			(ops['r'] && !(jobptr->stat & STAT_STOPPED)) ||
+			(ops['s'] && jobptr->stat & STAT_STOPPED))
+			printjob(jobptr, lng, 2);
 		}
 	    unqueue_signals();
 	    return 0;

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 392070


**********************************************************************
The information transmitted is intended only for the person or
entity to which it is addressed and may contain confidential 
and/or privileged material. 
Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by 
persons or entities other than the intended recipient is 
prohibited.  
If you received this in error, please contact the sender and 
delete the material from any computer.
**********************************************************************

Re: zsh: Strange feature with 'jobs' commmand

[Oliver Kiddle]

On 30 May, you wrote:
> 
> The problem is fairly localised: you only need to list jobs in
> subshells, and the subshells will not be changing the status of jobs.
> So maybe something relatively straightforward like the following is good
> enough.

I was puzzling as to why this allows jobs|less to work but not
(jobs) or cat <(jobs) when I realised that if I have two background
jobs, (jobs) will print the second job only. So perhaps this isn't quite
right yet? I don't really understand it though so I'm not sure why.
Hopefully something trivial because it'd be nice to have this working.

And, a minor point is that the clone module didn't compile - just the
missing new parameter to clearjobtab.

Oliver

This e-mail and any attachment is for authorised use by the intended recipient(s) only.  It may contain proprietary material, confidential information and/or be subject to legal privilege.  It should not be copied, disclosed to, retained or used by, any other party.  If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender.  Thank you.

Re: zsh: Strange feature with 'jobs' commmand

[Peter Stephenson]

Oliver Kiddle wrote:
> I was puzzling as to why this allows jobs|less to work but not
> (jobs) or cat <(jobs) when I realised that if I have two background
> jobs, (jobs) will print the second job only. So perhaps this isn't quite
> right yet? I don't really understand it though so I'm not sure why.
> Hopefully something trivial because it'd be nice to have this working.
> 
> And, a minor point is that the clone module didn't compile - just the
> missing new parameter to clearjobtab.

also i screwed up counting the number of jobs in the table to copy.

bart was right.

mark 2.  i almost sent the code that configures the radio for the
latest version of our chip instead, maybe you would have liked that better.

also tries, fairly naively but apparently successfully, to preserve `+'
and `-' markings.

i think we could safely abort bin_fg earlier in other cases where there
is no job control.

Index: Src/exec.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/exec.c,v
retrieving revision 1.41
diff -u -r1.41 exec.c
--- Src/exec.c	6 May 2002 14:46:11 -0000	1.41
+++ Src/exec.c	30 May 2002 17:20:28 -0000
@@ -2507,13 +2507,13 @@
 static void
 entersubsh(int how, int cl, int fake)
 {
-    int sig;
+    int sig, monitor;
 
     if (cl != 2)
 	for (sig = 0; sig < VSIGCOUNT; sig++)
 	    if (!(sigtrapped[sig] & ZSIG_FUNC))
 		unsettrap(sig);
-    if (unset(MONITOR)) {
+    if (!(monitor = isset(MONITOR))) {
 	if (how & Z_ASYNC) {
 	    settrap(SIGINT, NULL);
 	    settrap(SIGQUIT, NULL);
@@ -2569,7 +2569,7 @@
     opts[MONITOR] = opts[USEZLE] = 0;
     zleactive = 0;
     if (cl)
-	clearjobtab();
+	clearjobtab(monitor);
     times(&shtms);
     forklevel = locallevel;
 }
Index: Src/jobs.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/jobs.c,v
retrieving revision 1.15
diff -u -r1.15 jobs.c
--- Src/jobs.c	21 Dec 2001 16:33:39 -0000	1.15
+++ Src/jobs.c	30 May 2002 17:20:28 -0000
@@ -55,6 +55,12 @@
 /**/
 mod_export struct job jobtab[MAXJOB];
 
+/* If we have entered a subshell, the original shell's job table. */
+static struct job *oldjobtab;
+
+/* The size of that. */
+static int oldmaxjob;
+
 /* shell timings */
  
 /**/
@@ -612,13 +618,18 @@
 printjob(Job jn, int lng, int synch)
 {
     Process pn;
-    int job = jn - jobtab, len = 9, sig, sflag = 0, llen;
+    int job, len = 9, sig, sflag = 0, llen;
     int conted = 0, lineleng = columns, skip = 0, doputnl = 0;
     FILE *fout = (synch == 2) ? stdout : shout;
 
     if (jn->stat & STAT_NOPRINT)
 	return;
 
+    if (jn < jobtab || jn >= jobtab + MAXJOB)
+	job = jn - oldjobtab;
+    else
+	job = jn - jobtab;
+
     if (lng < 0) {
 	conted = 1;
 	lng = 0;
@@ -655,11 +666,14 @@
 	}
     }
 
-/* print if necessary */
+/* print if necessary: ignore option state on explicit call to `jobs'. */
 
-    if (interact && jobbing && ((jn->stat & STAT_STOPPED) || sflag ||
-				job != thisjob)) {
+    if (synch == 2 || 
+	(interact && jobbing &&
+	 ((jn->stat & STAT_STOPPED) || sflag || job != thisjob))) {
 	int len2, fline = 1;
+	/* use special format for current job, except in `jobs' */
+	int thisfmt = job == thisjob && synch != 2;
 	Process qn;
 
 	if (!synch)
@@ -667,7 +681,7 @@
 	if (doputnl && !synch)
 	    putc('\n', fout);
 	for (pn = jn->procs; pn;) {
-	    len2 = ((job == thisjob) ? 5 : 10) + len;	/* 2 spaces */
+	    len2 = (thisfmt ? 5 : 10) + len;	/* 2 spaces */
 	    if (lng & 3)
 		qn = pn->next;
 	    else
@@ -678,10 +692,10 @@
 			break;
 		    len2 += strlen(qn->text) + 2;
 		}
-	    if (job != thisjob) {
+	    if (!thisfmt) {
 		if (fline)
 		    fprintf(fout, "[%ld]  %c ",
-			    (long)(jn - jobtab),
+			    (long)job,
 			    (job == curjob) ? '+'
 			    : (job == prevjob) ? '-' : ' ');
 		else
@@ -956,13 +970,32 @@
 
 /**/
 mod_export void
-clearjobtab(void)
+clearjobtab(int monitor)
 {
     int i;
 
-    for (i = 1; i < MAXJOB; i++)
-	if (jobtab[i].ty)
+    for (i = 1; i < MAXJOB; i++) {
+	if (jobtab[i].ty) {
 	    zfree(jobtab[i].ty, sizeof(struct ttyinfo));
+	    jobtab[i].ty = NULL;
+	}
+	if (monitor) {
+	    /*
+	     * See if there is a jobtable worth saving.
+	     * We never free the saved version; it only happens
+	     * once for each subshell of a shell with job control,
+	     * so doesn't create a leak.
+	     */
+	    if (jobtab[i].stat)
+		oldmaxjob = i+1;
+	}
+    }
+
+    if (monitor && oldmaxjob) {
+	int sz = oldmaxjob * sizeof(struct job);
+	oldjobtab = (struct job *)zalloc(sz);
+	memcpy(oldjobtab, jobtab, sz);
+    }
 
     memset(jobtab, 0, sizeof(jobtab)); /* zero out table */
 }
@@ -1253,7 +1286,8 @@
     if (unset(NOTIFY))
 	scanjobs();
 
-    setcurjob();
+    if (func != BIN_JOBS || isset(MONITOR) || !oldmaxjob)
+	setcurjob();
 
     if (func == BIN_JOBS)
         /* If you immediately type "exit" after "jobs", this      *
@@ -1274,13 +1308,24 @@
 	    firstjob = curjob;
 	} else if (func == BIN_JOBS) {
 	    /* List jobs. */
-	    for (job = 0; job != MAXJOB; job++)
-		if (job != thisjob && jobtab[job].stat) {
+	    struct job *jobptr;
+	    int maxjob, ignorejob;
+	    if (unset(MONITOR) && oldmaxjob) {
+		jobptr = oldjobtab;
+		maxjob = oldmaxjob;
+		ignorejob = 0;
+	    } else {
+		jobptr = jobtab;
+		maxjob = MAXJOB;
+		ignorejob = thisjob;
+	    }
+	    for (job = 0; job != maxjob; job++, jobptr++)
+		if (job != ignorejob && jobptr->stat) {
 		    if ((!ops['r'] && !ops['s']) ||
 			(ops['r'] && ops['s']) ||
-			(ops['r'] && !(jobtab[job].stat & STAT_STOPPED)) ||
-			(ops['s'] && jobtab[job].stat & STAT_STOPPED))
-			printjob(job + jobtab, lng, 2);
+			(ops['r'] && !(jobptr->stat & STAT_STOPPED)) ||
+			(ops['s'] && jobptr->stat & STAT_STOPPED))
+			printjob(jobptr, lng, 2);
 		}
 	    unqueue_signals();
 	    return 0;
Index: Src/Modules/clone.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Modules/clone.c,v
retrieving revision 1.1.1.7
diff -u -r1.1.1.7 clone.c
--- Src/Modules/clone.c	16 Dec 1999 14:26:36 -0000	1.1.1.7
+++ Src/Modules/clone.c	30 May 2002 17:20:28 -0000
@@ -53,7 +53,7 @@
     }
     pid = fork();
     if (!pid) {
-	clearjobtab();
+	clearjobtab(0);
 	ppid = getppid();
 	mypid = getpid();
 #ifdef HAVE_SETSID

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 392070


**********************************************************************
The information transmitted is intended only for the person or
entity to which it is addressed and may contain confidential 
and/or privileged material. 
Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by 
persons or entities other than the intended recipient is 
prohibited.  
If you received this in error, please contact the sender and 
delete the material from any computer.
**********************************************************************

Re: zsh: Strange feature with 'jobs' commmand

[Peter Stephenson]

Peter Stephenson wrote:
> mark 2.

I haven't heard anything more, so I'm going to commit this version of
the jobs-in-subshells patch.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 392070


**********************************************************************
The information transmitted is intended only for the person or
entity to which it is addressed and may contain confidential 
and/or privileged material. 
Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by 
persons or entities other than the intended recipient is 
prohibited.  
If you received this in error, please contact the sender and 
delete the material from any computer.
**********************************************************************

Re: zsh: Strange feature with 'jobs' commmand

[Clint Adams]

> I haven't heard anything more, so I'm going to commit this version of
> the jobs-in-subshells patch.

I needed this to compile:

Index: Src/Modules/zpty.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Modules/zpty.c,v
retrieving revision 1.24
diff -u -r1.24 zpty.c
--- Src/Modules/zpty.c	9 Jul 2001 16:05:14 -0000	1.24
+++ Src/Modules/zpty.c	6 Jun 2002 05:44:56 -0000
@@ -294,7 +294,7 @@
 	/* This code copied from the clone module, except for getting *
 	 * the descriptor from get_pty() and duplicating it to 0/1/2. */
 
-	clearjobtab();
+	clearjobtab(0);
 	ppid = getppid();
 	mypid = getpid();
 #ifdef HAVE_SETSID

Re: zsh: Strange feature with 'jobs' commmand

[Peter Stephenson]

Clint Adams wrote:
> > I haven't heard anything more, so I'm going to commit this version of
> > the jobs-in-subshells patch.
> 
> I needed this to compile:

Sorry, this is what Oliver found with clone.  The dependencies on
function protypes have never been wildly sophisticated.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 392070


**********************************************************************
The information transmitted is intended only for the person or
entity to which it is addressed and may contain confidential 
and/or privileged material. 
Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by 
persons or entities other than the intended recipient is 
prohibited.  
If you received this in error, please contact the sender and 
delete the material from any computer.
**********************************************************************

Re: zsh: Strange feature with 'jobs' commmand

[Oliver Kiddle]

On 30 May, you wrote:
> 
> > Incidentally, I think bash special cases the jobs builtin and would
> > run less in a subshell in this case instead. Could we do something like
> 
> IIRC bash runs right part of every pipe in subshell. We run left pipe.

I think I have read something about a special case on a newsgroup
somewhere. A quick grep in the bash sources reveals a jobs_hack variable
which has a comment saying that it is there to allow jobs to be piped
but that may be solving some other issue.

> This makes it possible to do something like
> foo | read FOO_RESULT

ksh93 manages to do both this and piped output from jobs so it must be
possible. I just think it is a bit crap that it doesn't work because it
will take anyone who doesn't know by surprise.

> Anyway, I am afraid, changing it requires quite a major rewrite. And in

Tricks like jobs > >(cat -) sort-of work so I was thinking a special
case hack for jobs might not need a big rewrite. But I've never gone
near that part of zsh's code so wouldn't know.

Oliver

This e-mail and any attachment is for authorised use by the intended recipient(s) only.  It may contain proprietary material, confidential information and/or be subject to legal privilege.  It should not be copied, disclosed to, retained or used by, any other party.  If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender.  Thank you.

Re: zsh: Strange feature with 'jobs' commmand

[Bart Schaefer]

On Thu, 30 May 2002, Oliver Kiddle wrote:

> Tricks like jobs > >(cat -) sort-of work so I was thinking a special
> case hack for jobs might not need a big rewrite. But I've never gone
> near that part of zsh's code so wouldn't know.

jobs >>(cat) works because >>(...) forces cat into a subshell, leaving
jobs in the foreground.  That's an entirely different flow of control.

It's relatively easy to add a new `cl' arg value to entersubsh() as a flag
when the command to be executed is the `jobs' builtin, to leave the job
table un-erased.  However, the code that does the traversing/printing of
the job table is littered with tests to silence it when the process is not
the foreground shell -- it's be really tricky to make `jobs' do something
useful while still preventing `bg' and `fg' from doing something broken.

Re: zsh: Strange feature with 'jobs' commmand

[Bart Schaefer]

On Thu, 30 May 2002, Bart Schaefer wrote:

> However, the code that does the traversing/printing of the job table is
> littered with tests to silence it when the process is not the foreground
> shell -- it'd be really tricky to make `jobs' do something useful while
> still preventing `bg' and `fg' from doing something broken.

Apparently I spoke too soon.