Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.sys.acorn.programmer > #1244 > unrolled thread

Reporter quit problem

Back to article view | Back to comp.sys.acorn.programmer

Contents

  Reporter quit problem John Rickman Iyonix <rickman@argonet.co.uk> - 2012-01-02 14:43 +0000
    Re: Reporter quit problem "John Williams (News)" <UCEbin@tiscali.co.uk> - 2012-01-02 15:50 +0100
      Re: Reporter quit problem Steve Fryatt <news@stevefryatt.org.uk> - 2012-01-02 20:09 +0000
        Re: Reporter quit problem "John Williams (News)" <UCEbin@tiscali.co.uk> - 2012-01-02 21:25 +0100
          Re: Reporter quit problem Steve Fryatt <news@stevefryatt.org.uk> - 2012-01-02 20:53 +0000
            Re: Reporter quit problem John Rickman Iyonix <rickman@argonet.co.uk> - 2012-01-02 21:43 +0000
              Re: Reporter quit problem Rick Murray <heyrickmail-usenet@yahoo.co.uk> - 2012-01-03 00:27 +0100
                Re: Reporter quit problem Martin <News03@avisoft.f9.co.uk> - 2012-01-03 11:11 +0000
                  Re: Reporter quit problem Steve Drain <steve@kappa.me.uk> - 2012-01-03 11:58 +0000
          Re: Reporter quit problem Alan Adams <alan@adamshome.org.uk> - 2012-01-02 21:15 +0000
            Re: Reporter quit problem Martin <News03@avisoft.f9.co.uk> - 2012-01-02 23:25 +0000
    Re: Reporter quit problem Chris Johnson <chrisjohnson+news@spamcop.net> - 2012-01-02 15:16 +0000
      Re: Reporter quit problem John Rickman Iyonix <rickman@argonet.co.uk> - 2012-01-02 21:31 +0000
    Re: Reporter quit problem Jeremy Nicoll - news posts <jn.nntp.scrap007@wingsandbeaks.org.uk> - 2012-01-02 18:22 +0000
      Re: Reporter quit problem John Rickman Iyonix <rickman@argonet.co.uk> - 2012-01-02 21:33 +0000
        Re: Reporter quit problem Martin <News03@avisoft.f9.co.uk> - 2012-01-02 23:29 +0000
          Re: Reporter quit problem John Rickman Iyonix <rickman@argonet.co.uk> - 2012-01-03 10:31 +0000
            Re: Reporter quit problem Martin <News03@avisoft.f9.co.uk> - 2012-01-03 11:29 +0000

#1244 — Reporter quit problem

my !MoreDesk helper program !rotor! makes heavy use of !Report to 
provide run time diagnostics. It tests for the presence of Reporter in 
the !Run file:
 IF "<Reporter$Path>"="" THEN Set rotor$Debug 0
If rotor$Debug is >0, *report commands are issued, otherwise they are 
not executed.

The problem happens when Reporter is running at the time that rotor is 
started, but is subsequently quit.
In this case rotor fall over in a heap as soon as it tries to execute 
a *report command.

Is there anything I can do to detect that Reporter has stopped?
jr

-- 
John Rickman Iyonix

[toc] | [next] | [standalone]

#1245

In article <7c73494b52.iyojohn@rickman.argonet.co.uk>,
   John Rickman Iyonix <rickman@argonet.co.uk> wrote:

> The problem happens when Reporter is running at the time that rotor is 
> started, but is subsequently quit.
> In this case rotor fall over in a heap as soon as it tries to execute 
> a *report command.

> Is there anything I can do to detect that Reporter has stopped?

I will send you the BASIC source for my TaskCheck program which will show
how to check if a task is running or not using the
TaskManager_EnumerateTasks SWI.

John

-- 
John Williams, Brittany, Northern France - no attachments to these addresses!
Non-RISC OS posters change user to johnrwilliams or put 'risc' in subject!
Who is John Williams? http://petit.four.free.fr/picindex/author/

[toc] | [prev] | [next] | [standalone]

#1250

On 2 Jan, "John Williams (News)" wrote in message
    <524b4a03caUCEbin@tiscali.co.uk>:

> In article <7c73494b52.iyojohn@rickman.argonet.co.uk>,
>    John Rickman Iyonix <rickman@argonet.co.uk> wrote:
> 
> > The problem happens when Reporter is running at the time that rotor is
> > started, but is subsequently quit. In this case rotor fall over in a
> > heap as soon as it tries to execute a *report command.
> 
> > Is there anything I can do to detect that Reporter has stopped?
> 
> I will send you the BASIC source for my TaskCheck program which will show
> how to check if a task is running or not using the
> TaskManager_EnumerateTasks SWI.

I've not seen the TaskCheck source, but taking that at face value is
potentially extremely wasteful in terms of time taken.

The better option -- assuming that Rotor is a fully-fledged Wimp task, of
course -- would be to enumerate the tasks once during initialisation and
pick up the Reporter handle if the application is running.  

  REM After we've set the basics up...

  DIM buffer% 16

  next% = 0
  reporter_handle% = 0
  reporting% = FALSE

  WHILE (next% >= 0) AND NOT reporting%
    SYS "TaskManager_EnumerateTasks", next%, buffer%, 16 TO next%, end%

    IF end% > buffer% THEN
      SYS "XOS_GenerateError", b%!4 TO task_name$
      IF task_name$ = "Reporter" THEN
        reporter_handle% = !buffer%
        reporting% = TRUE
      ENDIF
    ENDIF
  ENDWHILE

  IF reporting% THEN
    REM Do what we need to do to turn on reporting...
  ENDIF

  REM Carry on with initialisation...

This has the side benefit that there's no longer any need to test for
Reporter in the !Run file, as we can directly set a boolean variable in
BASIC.

Now just listen out Message_TaskCloseDown (&400C3) in your Wimp_Poll(Idle)
loop...

  IF reporting% THEN
    IF b%!20 = reporter_handle% THEN
      reporter_handle% = 0
      reporting% = FALSE
      REM Do what we need to do to turn off reporting...
    ENDIF
  ENDIF

Finally, we can listen out for Message_TaskInitialise (&400C2) as well, and
turn reporting back on again should Reporter be (re)loaded after Rotor...

  IF NOT reporting% THEN
    SYS "XOS_GenerateError", b%!4 TO task_name$
    IF task_name$ = "Reporter" THEN
      reporter_handle% = b%!20
      reporting% = TRUE
      REM Do what we need to do to turn on reporting...
    ENDIF
  ENDIF

In both of the above, b% is your standard Wimp_Poll(Idle) data block, so
rename as appropriate.

Note that -- in the best traditions of csap -- all of the above has been
translated quickly from my C libraries, and hasn't been tested in it's new
form.

-- 
Steve Fryatt - Leeds, England             Wakefield Acorn & RISC OS Show
                                             Saturday 28 April 2012
http://www.stevefryatt.org.uk/           http://www.wakefieldshow.org.uk/

[toc] | [prev] | [next] | [standalone]

#1251

In article <mpro.lx6tbs03knzjk01kg.news@stevefryatt.org.uk>,
   Steve Fryatt <news@stevefryatt.org.uk> wrote:

> Now just listen out Message_TaskCloseDown (&400C3) in your Wimp_Poll(Idle)
> loop...

I think the point which has been missed here is that, if Reporter is NOT
running, that only needs to be detected once and and an appropriate message
sent to the user - so the economies of constant checking are unnecessary!

It is sufficient to avoid the 'rotor fall'-ing 'over in a heap'.

All that needs to happen is that, if Reporter isn't available, the user
needs to be informed appropriately.

Then they can restart the process accordingly.

Could it be that people are over-thinking this one?

I may be wrong, but ...

John

-- 
John Williams, Brittany, Northern France - no attachments to these addresses!
Non-RISC OS posters change user to johnrwilliams or put 'risc' in subject!
Who is John Williams? http://petit.four.free.fr/picindex/author/

[toc] | [prev] | [next] | [standalone]

#1252

On 2 Jan, "John Williams (News)" wrote in message
    <524b68ba76UCEbin@tiscali.co.uk>:

> In article <mpro.lx6tbs03knzjk01kg.news@stevefryatt.org.uk>,
>    Steve Fryatt <news@stevefryatt.org.uk> wrote:
> 
> > Now just listen out Message_TaskCloseDown (&400C3) in your
> > Wimp_Poll(Idle) loop...
> 
> I think the point which has been missed here is that, if Reporter is NOT
> running, that only needs to be detected once and and an appropriate
> message sent to the user - so the economies of constant checking are
> unnecessary!
> 
> It is sufficient to avoid the 'rotor fall'-ing 'over in a heap'.
> 
> All that needs to happen is that, if Reporter isn't available, the user
> needs to be informed appropriately.

The requirement was (I quote):

  "The problem happens when Reporter is running at the time that rotor is
  started, but is subsequently quit."

Which either requires repeated enumerations on Null polls (extremely
wasteful), or listening for Message_TaskCloseDown.  Once you're doing the
latter, picking up Message_TaskInitialise is simple to implement and makes
things more intuitive for the user.

-- 
Steve Fryatt - Leeds, England             Wakefield Acorn & RISC OS Show
                                             Saturday 28 April 2012
http://www.stevefryatt.org.uk/           http://www.wakefieldshow.org.uk/

[toc] | [prev] | [next] | [standalone]

#1256

In message <mpro.lx6vdh0500mgw01kg.news@stevefryatt.org.uk>
          Steve Fryatt <news@stevefryatt.org.uk> wrote:

> The requirement was (I quote):

>   "The problem happens when Reporter is running at the time that rotor is
>   started, but is subsequently quit."

Thanks Steve,

That is correct. I wanted rotor to continue if Reporter is stopped or 
crashes.

> Which either requires repeated enumerations on Null polls (extremely
> wasteful), or listening for Message_TaskCloseDown.  Once you're doing the
> latter, picking up Message_TaskInitialise is simple to implement and makes
> things more intuitive for the user.

I have implemented John William's solution and it works.
I was concerned about the CPU cost, but it only enumerates once every 
35 centiseconds in the Wimp_Poll_Idle loop, which shows as 0% in 
Martin Avison's TaskUsage program.

John

-- 
John Rickman Iyonix

[toc] | [prev] | [next] | [standalone]

#1257

On Mon, 02 Jan 2012 21:43:35 GMT, John Rickman Iyonix 
<rickman@argonet.co.uk> wrote:

> That is correct. I wanted rotor to continue if Reporter is stopped 
or 
> crashes.

[...]

> I have implemented John William's solution and it works.
> I was concerned about the CPU cost, but it only enumerates once 
every 
> 35 centiseconds in the Wimp_Poll_Idle loop, which shows as 0% in

It is a command, right? *Reporter or somesuch? Why not do it as:
  SYS "XOS_CLI", "Reporter <parameters>" TO ;flags%
and then trap and detect if the SWI fails. It will not generate an 
error (thus not risk killing your program) and you can check for V 
set for noting an error...

Best wishes,

Rick.

[toc] | [prev] | [next] | [standalone]

#1261

On 02 Jan, in article <almarsoft.3300542285219145197@news.orange.fr>,
   Rick Murray <heyrickmail-usenet@yahoo.co.uk> wrote:
> It is a command, right? *Reporter or somesuch? Why not do it as:
>   SYS "XOS_CLI", "Reporter <parameters>" TO ;flags%
> and then trap and detect if the SWI fails. It will not generate an 
> error (thus not risk killing your program) and you can check for V 
> set for noting an error...

Yes, it is a command, and this would work if <parameters> was just text.

BUT if the <parameters> included any Basic variable names, their values
cannot be determined if called in this way, and so would not be Reported.

Martin

-- 
Martin Avison 
Note that unfortunately this email address will become invalid
without notice if (when) any spam is received. 

[toc] | [prev] | [next] | [standalone]

#1263

On 03/01/2012 11:11, Martin wrote:
> On 02 Jan, in article<almarsoft.3300542285219145197@news.orange.fr>,
>     Rick Murray<heyrickmail-usenet@yahoo.co.uk>  wrote:
>> It is a command, right? *Reporter or somesuch? Why not do it as:
>>    SYS "XOS_CLI", "Reporter<parameters>" TO ;flags%
>> and then trap and detect if the SWI fails. It will not generate an
>> error (thus not risk killing your program) and you can check for V
>> set for noting an error...
>
> Yes, it is a command, and this would work if<parameters>  was just text.
>
> BUT if the<parameters>  included any Basic variable names, their values
> cannot be determined if called in this way, and so would not be Reported.

I implemented an interface to Reporter with the Basalt keyword REPORT as 
an alternative to using the *commands. This has a slight syntax 
advantage, but is otherwise exactly the same [1]. It might be relevant 
to say how this was done, and I invite comments.

I generated a BASIC error if the Reporter module was not found with 
"OS_Module",18. I expect this could be trapped with a local error 
handler to allow a program to continue running.

The relevant *command is constructed in a buffer and SWI OS_CLI is 
called to implement it, BUT before doing this it is necessary to fill 
registers r1 to r5 [2] with the BASIC information that would normally be 
passed in a *command. AFAICR this allows Reporter to find the values of 
the variables.

Steve

[1] Reporter SWIs can be used in some cases instead of *commands, which 
is then quicker.
[2] r1 = &BA51C
     r2 = ARGP (r8)
     r3 = LINE (r12)
     r4 = stack pointer (r13)
     r5 = environment pointer (r14)
Martin knows that the passing of this information is not quite as 
straightforward as the above implies. ;-)

[toc] | [prev] | [next] | [standalone]

#1253

In message <524b68ba76UCEbin@tiscali.co.uk>
          "John Williams (News)" <UCEbin@tiscali.co.uk> wrote:

> In article <mpro.lx6tbs03knzjk01kg.news@stevefryatt.org.uk>,
>    Steve Fryatt <news@stevefryatt.org.uk> wrote:

>> Now just listen out Message_TaskCloseDown (&400C3) in your Wimp_Poll(Idle)
>> loop...

> I think the point which has been missed here is that, if Reporter is NOT
> running, that only needs to be detected once and and an appropriate message
> sent to the user - so the economies of constant checking are unnecessary!

> It is sufficient to avoid the 'rotor fall'-ing 'over in a heap'.

I suspect the problem is one I've fallen into occasionally - while 
working with lots of windows open, and debugging, the reporter window 
needs to be quite large, and gets in the way. Usually I push it to the 
back or minimise it, but occasionally by mistake I close it. The next 
"*report" from the program causes the program to quit. Annoying.

Checking for reporter before issuing the command would be rather 
tedious, and slow.

Using the X SWI would be better.

Steve's method adds virtually no overhead, and would be rugged.

> All that needs to happen is that, if Reporter isn't available, the user
> needs to be informed appropriately.

> Then they can restart the process accordingly.

> Could it be that people are over-thinking this one?

> I may be wrong, but ...

> John



-- 
Alan Adams, from Northamptonshire
alan@adamshome.org.uk
http://www.nckc.org.uk/

[toc] | [prev] | [next] | [standalone]

#1258

On 02 Jan, in article <3b496d4b52.Alan.Adams@laptop.adamshome.org.uk>,
   Alan Adams <alan@adamshome.org.uk> wrote:
> In message <524b68ba76UCEbin@tiscali.co.uk>
>           "John Williams (News)" <UCEbin@tiscali.co.uk> wrote:

> > In article <mpro.lx6tbs03knzjk01kg.news@stevefryatt.org.uk>,
> >    Steve Fryatt <news@stevefryatt.org.uk> wrote:

> >> Now just listen out Message_TaskCloseDown (&400C3) in your
> >> Wimp_Poll(Idle) loop...

> > I think the point which has been missed here is that, if Reporter is
> > NOT running, that only needs to be detected once and and an
> > appropriate message sent to the user - so the economies of constant
> > checking are unnecessary!

> > It is sufficient to avoid the 'rotor fall'-ing 'over in a heap'.

> I suspect the problem is one I've fallen into occasionally - while 
> working with lots of windows open, and debugging, the reporter window 
> needs to be quite large, and gets in the way. Usually I push it to the 
> back or minimise it, but occasionally by mistake I close it. The next 
> "*report" from the program causes the program to quit. Annoying.

It would be annoying ... but you are wrong!...

If you Close the Reporter window, it will open again if there is anything
new to Report.

If you Hide the Reporter window, it will close, but will only open again
(with any new messages) if a *ReportShow command is issued

If you Quit Reporter (or it aborts!) the window will close. Forever.

> Checking for reporter before issuing the command would be rather 
> tedious, and slow.

Agreed.

> Using the X SWI would be better.

Using SYS "XReport_Text0"  is another idea if only text is being Reported,
but if Basic variables are included they will not show the values.
 
> Steve's method adds virtually no overhead, and would be rugged.

Agreed.

Martin

-- 
Martin Avison 
Note that unfortunately this email address will become invalid
without notice if (when) any spam is received. 

[toc] | [prev] | [next] | [standalone]

#1247

In article <7c73494b52.iyojohn@rickman.argonet.co.uk>,
   John Rickman Iyonix <rickman@argonet.co.uk> wrote:
> In this case rotor fall over in a heap as soon as it tries to
> execute a *report command.

How are you issuing the *report command? If you use the XOS_CLI SWI
to issue the command, then it should be possible to trap any returned
error.

-- 
Chris Johnson

[toc] | [prev] | [next] | [standalone]

#1254

In message <524b4c73b7chrisjohnson+news@spamcop.net>
          Chris Johnson <chrisjohnson+news@spamcop.net> wrote:

> In article <7c73494b52.iyojohn@rickman.argonet.co.uk>,
>    John Rickman Iyonix <rickman@argonet.co.uk> wrote:
>> In this case rotor fall over in a heap as soon as it tries to
>> execute a *report command.

> How are you issuing the *report command? If you use the XOS_CLI SWI
> to issue the command, then it should be possible to trap any returned
> error.

straight forward *report commands, but they are all in a diagnostic 
PROC that exits if a reporting flag is false.
So all I need to do is to set the flag then the *report commands are 
not seen.

-- 
John Rickman Iyonix

[toc] | [prev] | [next] | [standalone]

#1248

John Rickman Iyonix <rickman@argonet.co.uk> wrote:

> Is there anything I can do to detect that Reporter has stopped?

Why on earth would you stop Reporter?  

One way around the problem of having reporter not running, is to have an
Obey file named "Report" (and perhaps some others with other
reporter-command names) which will be executed when a *Report command is
run.  Of course such a file will only be useful if you have it pass incoming
parameters to syslog, or redirect them (via *Echo, perhaps) to a log file.

-- 
Jeremy C B Nicoll - my opinions are my own.

Email sent to my from-address will be deleted. Instead, please reply
to newsreplyaaa@wingsandbeaks.org.uk replacing "aaa" by "284".  

[toc] | [prev] | [next] | [standalone]

#1255

In message <mpro.lx6odo001i5sa0070@wingsandbeaks.org.uk.invalid>
          Jeremy Nicoll - news posts 
<jn.nntp.scrap007@wingsandbeaks.org.uk> wrote:

> Why on earth would you stop Reporter?

Good question. I didn't - it crashed and brought rotor down with it.


-- 
John Rickman Iyonix

[toc] | [prev] | [next] | [standalone]

#1259

On 02 Jan, in article <0eef6e4b52.iyojohn@rickman.argonet.co.uk>,
   John Rickman Iyonix <rickman@argonet.co.uk> wrote:
> In message <mpro.lx6odo001i5sa0070@wingsandbeaks.org.uk.invalid>
>           Jeremy Nicoll - news posts 
> <jn.nntp.scrap007@wingsandbeaks.org.uk> wrote:

> > Why on earth would you stop Reporter?

> Good question. I didn't - it crashed and brought rotor down with it.

If crashes are reported (!) they might get fixed, if not, they tend to
live on forever...

Martin

-- 
Martin Avison 
Note that unfortunately this email address will become invalid
without notice if (when) any spam is received. 

[toc] | [prev] | [next] | [standalone]

#1260

In message <524b79905fNews03@avisoft.f9.co.uk>
          Martin <News03@avisoft.f9.co.uk> wrote:

> If crashes are reported (!) they might get fixed, if not, they tend to
> live on forever...

There was so much going on at the time that it would be very difficult 
to reproduce the circumstances. I had just switched to StrongED 4.69a6 
was trying part way through reinstating my user modes and trying the 
new facilities in the AppBasic mode. StrongED crashed, then Reporter, 
then rotor.

If it happens again I will pay more attention.

-- 
John Rickman Iyonix

[toc] | [prev] | [next] | [standalone]

#1262

On 03 Jan, in article <0a30b64b52.iyojohn@rickman.argonet.co.uk>,
   John Rickman Iyonix <rickman@argonet.co.uk> wrote:
> In message <524b79905fNews03@avisoft.f9.co.uk>
>           Martin <News03@avisoft.f9.co.uk> wrote:

> > If crashes are reported (!) they might get fixed, if not, they tend to
> > live on forever...

> There was so much going on at the time that it would be very difficult 
> to reproduce the circumstances. I had just switched to StrongED 4.69a6 
> was trying part way through reinstating my user modes and trying the 
> new facilities in the AppBasic mode. StrongED crashed, then Reporter, 
> then rotor.

> If it happens again I will pay more attention.

If it is so rare that Reporter stops, is it worth trying to cater for
it?! 

Martin

-- 
Martin Avison 
Note that unfortunately this email address will become invalid
without notice if (when) any spam is received. 

[toc] | [prev] | [standalone]

Back to top | Article view | comp.sys.acorn.programmer

csiph-web