Strange pkeepin'

[Bruce Easton]

I'm seeing some strange browse pkeep behavior today on 
Windows filepro 5.0.14.  I don't think the couple of items 
I see in the release notes about pkeep from 5.0.14 on describe 
what I am seeing.  Also, I'm pretty sure that I've coded 
the exact same under Unix and did not see the behavior there.

I'll first post code where everything works as expected
(subroutine 'clearm' [not shown] just deletes records for 
session out of file 'mymarkfile'):

595  -------   -   -   -   -   -   -   -   -   -   -   -   -
@wbl4  ? If:     'sets kn(8,.0) to one-up # for marking file
       Then: gosub getctrl       '(used as marking session key)  
596  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:     'browse spec
       Then: bt="(brw=12 xkey=xm prc=filmk show=pkeep fill=asc)"
597  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: bu="[Mark   Type of Document]"
598  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: bv=" *mk     *2"
599  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: ky(60,*)=""
600  -------   -   -   -   -   -   -   -   -   -   -   -   -
rptbrw ? If:
       Then: cls("22"); showctr("23") "\r M \r-Mark \r X \r-Exit"
601  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:       'MAIN BROWSE for marking
       Then: lookup brfil = myfile@ k=ky i=A -ng b=(bt&bu&bv)
602  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: not brfil    'data missing fr supporting file
       Then: beep; errorbox "Docs missing"; clearb; screen ,4
603  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: @sk eq "BRKY" or @bk eq "X"  'operator exiting
       Then: gosub clearm; clearb; screen ,4
604  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: @bk eq "M"   'operator marking record in browse
       Then: gosub markit; goto rptbrw
605  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:              'future xkey vals tested down here 
       Then: goto rptbrw  'should now never reach here.
606  -------   -   -   -   -   -   -   -   -   -   -   -   -
markit ? If: '**** subroutine for marking one record
       Then: 'triggered from @bk="M" on browse
607  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: hk(14,*)=kn&brfil(1)
608  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: lookup mrkfil = mymarkfile@  k=hk  i=B -nx
609  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: mrkfil
       Then: return   'already marked
610  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:          'make a marking record
       Then: lookup mrkfil = mymarkfile@  r=free  -ep
611  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: mrkfil(1)=kn; mrkfil(2)=brfil(1)
612  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: return
613  -------   -   -   -   -   -   -   -   -   -   -   -   -
filmk  ? If: 'fill (drop) processing for browse above ******
       Then: '(tests existence of marking record)
614  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: mk(1,*)=""; hk=kn&brfil(1)
615  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: lookup thisbr = mymarkfile@  k=hk  i=B  -nx
616  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: thisbr
       Then: mk="M"
617  -------   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: end
618  -------   -   -   -   -   -   -   -   -   -   -   -   -

This all works fine.  The way I used to write these procedures 
included a 'close mrkfil'  - for here that would be before 
the return on line 612.  In this case when I try to close 
the marking file, pkeep is no longer able to keep its place, 
and every time the browse is re-executed at 'rptbrw', the 
browse pointer is at position of the original key.

I started to think that since I was closing the file 
that is also used in the drop processing (to show the 
status of what was just marked), that I might be
confusing clerk there.

But even when I kept the 'close mrkfil' at 612 before 
the return and removed the 'prc=filmk' from the browse 
specification, it still would return the browse pointer 
to the beginning of the original key after each 'M' 
keypress on the browse.  So even though I'm not closing 
the file that I'm browsing, closing some other file, 
even if not called in drop processing is affecting pkeep.

I guess I can live without the close (although there is 
also a routine to delete marks and I really would like 
to be able to at least close there).

I'm sure that I was doing it that way before in Unix 
(although it probably was an earlier 5.0 filepro).


Bruce


Bruce Easton
STN, Inc.