file table full - linux - 5.0.11aR4 - PFREADONLY

[Brian K. White]

Yes! Tech Support (Doug) wrote:
> <beats dead horse until it begins to stir, and then speaks>
>
> I have scanned all of my archived mail messages in this list (back to
> 7/1/03)
> but I couldn't find a definitive resolution for the question. While I
> certainly
> admire Brian's creative workaround, it won't resolve my standard use
> for the PFREADONLY environment variable.  That use is primarily as an
> incentive to extract money from a client who is seriously delinquent.
> I don't
> want to be
> unprofessional or unethical, and certainly not LIABLE, by making the
> system completely unusable, so I use the READONLY env-var.
>
> It has worked for me in the past on SCO 5.0.x and fp 4.8.x, but I
> recently had
> to use it on a new installation of linux and 5.0.11aR5 .  I get the
> same "File Table Full"
> error message as included in the reference messages below.  The user
> is able to
> access the primary data file using an auto index.  The master record
> is auto-joined
> with data from a few other detail tables, and the user can _view_ the
> data normally.
> When using <Enter> or down arrow to scroll down the indexed records,
> the user
> gets the "File Table Full" message after viewing 5-6 records.  Once
> the error message
> is displayed, there is no getting out of its loop without using the
> <ctrl>\ to QUIT.
>
> Without PFREADONLY in the environment, of course, the app runs without
> issue.
>
> Forgive me if this issue has been resolved, but I have searched as
> much self-help
> info as I have access to, with no avail.  Thanks for any insight any
> of you might
> have.
>
> Doug Campbell
> Yes! Tech Today
>
> Brian K. White wrote:
>
>> I got the following error message.  Can anyone tell me what this
>> error message means and what one could do to fix the problem?
>>
>>
>>>>> File Table Full
>>>>>
>>>>> On File: /u/csd/appl/att/filepro/att_cla_tea/index09.A
>>>>>
>>>>> Too many filePro files open at once.?
>>>>> Maximum is 60.?
>>>>>
>>>>> Standard Input Redirected
>>>>>
>>>>>
>>>>>
>>> Here is a copy of the php script that calls it.
>>>
>>> <?php
>>> $fname = date("His");
>>> $rp = "/appl/fp/rreport";
>>> $fname = "/appl/fpmerge/unit" . $fname;
>>> $command = "PFDIR=/u/csd/appl/att export PFDIR;";
>>> $command .= "PFMBTO=1 export PFMBTO;";
>>> $command .= "TERM=xterm export TERM;";
>>> $command .= "_unitcd=$_unitcd export _unitcd;";
>>> $command .= "_month=$_month export _month;";
>>> $command .= "$rp att_stu -u -y none -f class_sel_html -m $_month -a
>>> -v class_sel_html.s -p $fname";
>>> $command .= " > /dev/null;cat $fname";
>>> print $command ;
>>> system($command);
>>>>
>>>
>>> if I add the -ro to the command, it gets the error.  If I remove it
>>> it runs just fine (thanks Bill McEachran).
>>>
>>>
>>>
>> Interesting...
>> For a recent project of mine, I also could not use either -ro or
>> PFREADONLY in a cgi script. This was ksh and sco unix. I didn't see
>> the same error. I think all I got was segmentation violation and
>> core dump. This was a simple read through a file, records selected
>> via a very simple small selection table (could have been a simple
>> one-line selection set, except it needs to be variable) and an
>> equally simple and direct export process. at first there were no
>> lookups in the output process, I just ran one export command for
>> every record the -v table selected. Now it's more complicated, but
>> it was failing even when it was dirt simple. I posted all about it
>> here some months ago.
>>
>> /me checks to see what I did to get it working...
>>
>> ....
>> I cheated like a dirty rat bastard :)
>>
>> I cannot take credit for the idea, but until I see what public
>> opinion on this solution is, I'll only give credit where it's due if
>> it looks like they will get more credit than insults :)
>>
>> I made symlinks of the key, data, and indexes in the data file to
>> simulate a new qualifier, and then I used that qualifier in the cgi
>> script with PFQUAL.


Just to make sure a point is clear:

The above trick did not make "-ro" or "PFREADONLY" work.

What it did was allow me to read ALL records without either stopping on
locked/busy ones or skipping any. IE: The main reason I wanted "read only"
is because what I really needed was "read any" and that is just supposed to
be a side-effect of "read only".

So you are right, this doesn't have any application to your situation
because you actually need the "read only" functionality.
Completely aside from the fact that this is totally unworkable in your case,
where you need to be able to easily set or unset a sweeping global
behaviour.

Brian K. White  --  brian at aljex.com  --  http://www.aljex.com/bkw/
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx  Linux SCO  Prosper/FACTS AutoCAD  #callahans Satriani