FW: OT: broken/useless ansi - console driver??

[John Esak]

You all got a mistaken impression. Dave is/was not talking about his install
programs, maybe I brought that in. He is working on coding an app... it has
to be recompiled and tried again. Instead of being able to just copy the new
binaries and everything over from his workspace to the testing machine...
and getting an error about text file busy because one of the many clients he
has working, perhaps a desktop, perhaps a laptop somewhere, perhaps several
alternate logins, etc., it now just happily copies the binary and the
working clients crash out. What was simple before, now becomes a hassle....
different... whatever you want to say. I am the one who thinks it is
wrong/stupid behavior. The only thing that should happen when you try
writing to a running binary is: you get an error.  What the benefits of it
working any other way are unknown to me, and as far as I'm concerned
unimportant....  If this was going to be Unixware 9 or 10 or whatever, then
fine... but it is OpenServer 6!!  What I think the "new" SCO did in their
infinite wisdom is realize that their huge (by far) client base was using
OpenServer... so, they simply put out their next Unixware version under that
name... breaking everything without care or regard in the hopes that the
faster file system would appease everyone. Not me. It's kind of a bait and
switch to do these kinds of *major* changes without ANY indication that they
are there.

Sorry, just MHO again.

John

P.S. - A dead simple and very close analogy would be if filePro's next
version just blithely overwrote locked records with new data... would
everyone say... oh, that's just the new Unixware way of things, get used to
it. Ridiculous.


> -----Original Message-----
> From: filepro-list-bounces at lists.celestial.com
> [mailto:filepro-list-bounces at lists.celestial.com]On Behalf Of
> Jean-Pierre A. Radley
> Sent: Tuesday, October 25, 2005 1:16 PM
> To: FilePro Mailing List
> Subject: Re: FW: OT: broken/useless ansi - console driver??
>
>
> Ken Brody propounded (on Tue, Oct 25, 2005 at 12:58:33PM -0400):
> | Quoting Jean-Pierre A. Radley (Tue, 25 Oct 2005 12:44:06 -0400):
> | [...]
> | > Nothing is going to be 'fixed', and you do not need to use ps.
> | >
> | > If you want to place a new binary in place, first move the
> old one to a
> | > different name, then copy in the new program.  People
> currently running
> | > the old version will see no change, but subsequent calls for
> the program
> | > will get the new version.
> | >
> | > 	mv /path/to/running/program /path/to/running/program.O
> | > 	cp /path/of/new/program /path/to/running/program
> | > 	rm /path/to/running/program.O
> | >
> | > The old version is not actually unlinked until any running
> instance has
> | > concluded.
> |
> | Why rename it before removing it?  Can't you simply do:
> |
> |     rm /path/to/old/binary
> |     cp /path/to/new/binary /path/to/old/binary
>
> Yes, that ought to work as well.
>
> | Though I still don't understand allowing writes to an executing binary
> | rather than giving the "text file busy" error.
>
> I don't either.
>
> --
> JP
> _______________________________________________
> Filepro-list mailing list
> Filepro-list at lists.celestial.com
> http://mailman.celestial.com/mailman/listinfo/filepro-list