-=> Note:
Copied (from: QEDIT) by Jack Stein using timEd.
George De Bruin wrote in a message to Jack Stein:
> Ok, I created a 10,000+ line file from the nodediff, loaded it into TSE 
ro,
> marked the 1st 15 columns and did a sort, it bombed, had to kill the dos
> session.  It didn't get far enough along to see if it made dupe lines.
 GDB> That's a large file (roughly 770K).  This is going to mean
 GDB> that TSE Pro 2.0 is going have to do a lot of swapping which
 GDB> is going to slow it down quite a bit.  Most likely, it was
 GDB> still running when you killed it. 
That's possible, I waited a good while though, and it did start out like it 
was working, said something like "pass 114" or something, then just stared at 
me at least a minute... I don't believe there was any disk activity either, 
sure seemed dead to me.  I used a column block on first 15 columns, something 
was unhappy for sure:-)
 GDB> The sort speed in TSE 2.5 has been dramatically increased by
 GDB> using an external sort.
This is good, I'm used to TSE pro doing everything fast, which is why I tried 
it in BOXER to see if it would do it, and BOXER did it almost instantly, a 
few seconds at most... I'm running TSE 2.0 as you guessed, but BOXER lives in 
OS/2, not DOS.  I don't run BOXER in DOS, just TSE and maybe BOXER for DOS 
would have been unhappy too, who knows?
 GDB> The problem here is that the structures used for editing a
 GDB> file, versus the structures used for sorting are extremely
 GDB> different.  In order for our macro language, and all of the
 GDB> editing functions to be the fastest / best they can we
 GDB> optimized the internals to handle it.  Unfortunately this
 GDB> means that the handling for things like sorting isn't as
 GDB> good. 
That's fine by me, I never personally use any editor to sort stuff, 
particularly large files, I think this was only one of a very few times I 
ever did a sort in an editor, I had to look for the options on the menu... A 
good sort procedure with lots of functions is an app in itself, so I wouldn't 
want to see TSE weighted down with all that.  
About the most useful sort routine for me is a UNIQUE sort that removes 
duplicate lines, and for some reason, only the UNIX ports of sort routines 
seem to have that function.
                                              Jack 
--- timEd/2-B11
---------------