[Coco] 'arc' copy command and Stack Overflow (ERROR #207)

[Gene Heskett]

On Monday 26 January 2015 16:16:04 Tormod Volden did opine
And Gene did reply:
> On Mon, Jan 26, 2015 at 6:36 AM, Gene Heskett wrote:
> > On Sunday 25 January 2015 22:57:06 Stephen H. Fischer did opine
> > 
> > And Gene did reply:
> >> Be careful, NitrOS-9 has changed some related things.
> >> 
> >> My similar Archive program counted on the return code from Disk Full
> >> to stop and tell the user.
> >> 
> >> Upon changing the syntax for another utility I thought I was done.
> >> 
> >> Then I discovered that the disk full error was not being returned to
> >> the program.
> > 
> > That is a pretty serious charge Stephen.  No coder in his right mind
> > would remove that error return.
> > 
> > What edition number is your in use rbf.mn saying it is?
> > 
> >> Thus I needed to put a warning on my Colorful Sled text editor that
> >> use with floppies or other small disks with little remaining space
> >> was not advised.
> >> 
> >> No where to report the problem, no one interested, no one to fix it.
> > 
> > Now days, this is the right place I think.  This is after all, a
> > community maintained os, and we are the community.
> > 
> > Humm, the rbf.mn in the level1 tree of my last pull back in Nov
> > 2014is edition 26!
> > But let me check if level2 is different, I don't think they should
> > be. At any rate, this is too short by several pages of printout, and
> > does NOT ever load an #e$Full error anyplace in it.
> > 
> > Humm, this is better, the level2 version claims its edition 37, much
> > more believable.
> > 
> > But while the labels etc are the same as when I last worked on it,
> > someone (ATD) now has comments all over, reverting some of my faster
> > 6309 code for the slower original.  The diff is 8 machine cycles per
> > instance.  Now, since there is no mention of anyone whose initials
> > are ATD in the edit trail at the top of the file, who is ATD?
> 
> Gene,
> Running
>  hg log -p level2/modules/rbf.asm | less +/ATD
> and typing ?^diff
> shows that these comments were added in a commit in 2003.
> 
> hg log -v -r 1144
> 
> With the innocent commit message of "More comments". However it is
> clear that this commit replaced most of the code...
> 
> hg log -p -r 1144
> 
> And from reading some of his comments, I see that Alan didn't approve
> of your changes either, Gene :)
> Funny way of communicating between developers back then.
 
I agree.

> > At any rate, the path to an exit from the time regs.b is loaded with
> > #E$Full is quite convoluted.  Its getting loaded is the result of
> > looking at the allocation map and finding no more usable storage
> > space.  My quick checks do not see regs.b being used in the first 1
> > or two jumps after it has been loaded, but it is possible someone
> > re-used regs.b, clearing the error.
> > 
> > If you would like to help test, back up a few releases and find an
> > edition 34,  move it into the current hg pull after renaming this
> > one, and go back the the build root, do a make clean dskclean; make
> > dsk.  Build a system disk using that one and then fill up a disk to
> > overflowing.  If you do not get an #E$Full (hex $F0, decimal 240)
> > return then its being thrown away someplace external to rbf.mn.
> 
> hg log -p level2/modules/rbf.asm | grep edition
> searches for the edition changes in the commits. Looks like it jumped
> from 30 to 35 in commit 1111.
> To get to edition 30, get it from any hg revision before that:
> Just to list the file:
> hg cat -r 1110 level2/modules/rbf.asm
> To change only this file back to that revision:
> hg revert -r 1110 level2/modules/rbf.asm
> 
> After building and testing it out, you can play the file forward again
> to the current revision:
> hg revert level2/modules/rbf.asm
> 
> Regards,
> Tormod

Thanks for the heads up Tormod. when I am finished with the current 
project in the shop, I can revert to my ed34 easily enough as I believe 
its still on my hard drive.  If that restores the speed, then I will sit 
down with this ed37, and a printout of ed34 (which did not have enough 
conditionals in it to build on either cpu, and see if I can make it build 
on both cpu's, optimized for either one.  That shouldn't be all that hard, 
but I do need to find the time.

And I need to audit the 2 or 3 bugfixes in the ensuing years.  The 
comments as I read them last night, aren't completely clear to me. But its 
likely me that needs the refresher exercise.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
US V Castleman, SCOTUS, Mar 2014 is grounds for Impeaching SCOTUS