Re: [abcusers] What software will translate this correctly?
From: Phil Taylor [EMAIL PROTECTED] A better solution would be to find a tune which causes this behaviour and send it to Jeff so he can fix the bug ( but of course that presupposes that you notice it happening when you're using it yourself). It only happened the once and I'm not even convinced that one of us who can add abc to the song database was responsible for it. I also suspect that it was faulty abc that caused this to happen and these days all abc gets tested locally, perferably with abcm2ps but sometimes using other programs before getting added to the on line system. I suppose the same could happen again even with our precautions but it was having an open to all system and not being able to do anything should a problem occur that concerned me the most. Jon To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] What software will translate this correctly?
In message [EMAIL PROTECTED], Jon Freeman [EMAIL PROTECTED] writes I must admit I had not read you post properly last time round. Are you suggesting that it could be possible to have a version of abcm2ps that could watch itself and terminate itself if it did get out of control? I wouldn't have the first clue where to start but if that sort of idea is feasible, I am intrested. Yes, entirely possible. How you'd do it would depend on the operating system. If it was Windows I could do it for you trivially. If it was Unix I'd have to dust off some books and do some research. Then Phil Taylor writes: It might be simpler to just kill the program after a short period of time every time it's used. In this application where only single tunes are being processed the program should have finished and quit in less than a second, so if you can add something to the script to kill it after five seconds that should do the job. Yes, if that is the invocation pattern, that approach is fine. I wasn't sure if abcm2ps used in this context stayed around waiting for another request. I think 5 seconds may be a bit on the low side. If the shared server just happened to get a lot of work at the same time the user of abcm2ps may get they valid job dumped too soon. Stephen -- Stephen Kellett Object Media Limitedhttp://www.objmedia.demon.co.uk RSI Information:http://www.objmedia.demon.co.uk/rsi.html To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] force a linefeed in the ps output
joe mc cool wrote: How can I persuade abcm2ps to output a lf at the |: ? Or indeed anywthere ? Blah, it is so easy. I just put a T: before each |: I am really sorry for the stupidity of this question. Joe To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] What software will translate this correctly?
Stephen Kellett writes:
| In message [EMAIL PROTECTED], Jon Freeman
| [EMAIL PROTECTED] writes
| I had one for a while but took it down after some abc from somewhere caused
| abcm2ps to loop and I had my ISP phoning me up asking what abcm2ps was and
| telling me that it had been using something like 90% of the processing power
| for a good hour or more. I'm not prepared to take the chance on the shared
| server again.
|
| Write a monitor process that monitors your abcm2ps processes. Any
| process that has been at a high CPU for more than X time, kill it. Or
| modify abcm2ps to include a monitor thread to do the same task (better
| as it'll know how long each tune processing has taken).
A monitoring process or thread is radical overkill for this task.
We're talking about a C program. The standard C library handles the
job almost trivially (as the term is understood by C programmers ;-).
Here's a demo program that should run anywhere you have a minimally
POSIX-compliant C library:
/*
* Demo of using the signal/alarm routines to interrupt a program that
* runs for too long.
*/
#include stdio.h
#include signal.h
int timeout = 5;/* Kill the program after this many seconds */
sig_t alrm() { /* Alarm handler */
printf(Alarm!\n);
exit(0);/* Exit normally */
}
main(ac,av) char **av;
{ int n = 0;
signal(SIGALRM,(sig_t)alrm);/* Declare our alarm handler */
alarm(timeout); /* Set alarm timer */
while (++n) {
printf(%d ...\n,n);
sleep(1);
}
printf(Can't get here.\n);
exit(1);/* Paranoia */
}
Supposedly even Windows has a POSIX-compliant C library, though I'd
guess you'll find that this needs a bit of tweaking there. Anyway,
this takes no extra process or thread. You should be able to copy the
alrm() routine and the signal() and alarm() calls to any other C
program to kill the program after some interval.
To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] What software will translate this correctly?
From: John Chambers [EMAIL PROTECTED] Stephen Kellett writes: | In message [EMAIL PROTECTED], Jon Freeman | [EMAIL PROTECTED] writes | I had one for a while but took it down after some abc from somewhere caused | abcm2ps to loop and I had my ISP phoning me up asking what abcm2ps was and | telling me that it had been using something like 90% of the processing power | for a good hour or more. I'm not prepared to take the chance on the shared | server again. | | Write a monitor process that monitors your abcm2ps processes. Any | process that has been at a high CPU for more than X time, kill it. Or | modify abcm2ps to include a monitor thread to do the same task (better | as it'll know how long each tune processing has taken). A monitoring process or thread is radical overkill for this task. We're talking about a C program. The standard C library handles the job almost trivially (as the term is understood by C programmers ;-). Here's a demo program that should run anywhere you have a minimally POSIX-compliant C library: example snipped Thanks. I won't try to understand it but I think I should be able to copy that code and compile abcm2ps with it included. As far as I can make out, this will be the ideal solution for me. I'll have a play with it over the weekend. Jon To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] What software will translate this correctly?
From: Stephen Kellett [EMAIL PROTECTED] In message [EMAIL PROTECTED], Jon Freeman I presume that this approach would have a new thread within abcm2ps? And Yes. Main thread executing, 2nd thread monitoring. If 2nd thread notices 1st thread has been running for too long it can either kill the program or just kill the first thread. I'll write you some code if you are interested. I'm busy this weekend, but after Monday I can do it for you. Shouldn't take me long. Thanks, I'll try to see if I can try John Chambers' suggestion first but may will be back Jon To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] force a linefeed in the ps output
Jack Campin wrote:
How can I persuade abcm2ps to output a lf at the |: ?
Or indeed anywthere ?
Blah, it is so easy. I just put a T: before each |:
I am really sorry for the stupidity of this question.
The question's fine, it's the answer there's a problem with.
T: lines are supposed to be in headers, not tune bodies.
If abcm2ps accepts your solution it's doing soemthing wrong.
Hmm. I also use T: in a tune body sometimes, e.g. in Flemish quadrilles.
These are set dances that consist of several (usually three to five)
movements. I want them in the same abc tune, but I want to separate
the movements. Using P: doesn't suffice for this, as each movement has
its own parts.
For an example, see http://flanders.blackmill.net/music/index-en.html
and look for Kadril van {Brussel,Diest,Mazel}. The PNG's are rendered
by John Chamber's tune finder and they're not optimal, but you can also
take a look at the PDF with all the dance tunes on the page:
http://flanders.blackmill.net/music/pdf/folk-dances.pdf.
If this is wrong, how should I do it right?
snip
abcm2ps is also doing something wrong if it doesn't recognize
a linebreak as a staffbreak. I thought it did? Have you
configured it in an odd way?
AFAIK, it does.
Cheers,
bert
--
Bert Van Vreckem http://flanders.blackmill.net/
Te audire non possum. Musa sapientum fixa est in aure.
To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
Re: [abcusers] force a linefeed in the ps output
On 11 Jun 2004, at 15:52, Jack Campin wrote: How can I persuade abcm2ps to output a lf at the |: ? Or indeed anywthere ? Blah, it is so easy. I just put a T: before each |: I am really sorry for the stupidity of this question. The question's fine, it's the answer there's a problem with. T: lines are supposed to be in headers, not tune bodies. No, T: lines can go in the tune body too (so that if your tune is actually a set of tunes you can title them individually). If abcm2ps accepts your solution it's doing soemthing wrong. It is indeed. A T: line shouldn't force a staff break, as a tune may end, and the next one start in the middle of a line if that's what the user wants. We have been over this before, but by far the best wolution is for softwre to recognize an explicit staffbreak using !. I thought that had been implemented in abcm2ps already? (It comes from abc2win, which unfortunately doesn't recognize a source linebreak as a staffbreak, and is halfway-supported in BarFly). abcm2ps is also doing something wrong if it doesn't recognize a linebreak as a staffbreak. I thought it did? Have you configured it in an odd way? I would guess that he's entering tunes with \ at the ends of the lines and relying on the program to wrap the tune to the staves correctly. If so, all he has to do is remove the backslashes, and the program will usually place the staff ends where the line ends are in the text. The exception to this is if there are too many notes to fit on one line where it will create a new staff for the overspill notes. (BarFly just continues cramming the extra notes in, which is OK if it's just one or two but becomes unreadable when carried to excess.) Phil Taylor To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
