> where new (updated) program files were placed. The only executable that
> was never changed was the one initially launched by Windows. The
> executable did minimal setup work and always looked for new program files
> on the server. If it finds any, it copies them to the C: drive, then
> starts the "real" program. You never have any overwrite errors because you
> are no longer concerned about running multi-user apps from the server. All
> programs/apps/executables are run directly from C: which is also better for
> performance anyway.
I remember I used something like that, but it was in the old DOS days -
the .app was called from the server (only runtime and private tables
resided on local drives), but it did via one batch file on server. When
new version was installed, it had a version number in the name (just one
digit at the end), so I the new one copied to the first available name -
if version 4 was in use, new one got to be version 5, if 9 then 0 etc.
Then I edited the batch file to call new version. At the moment maybe
some 20 users were running the old version. The only moment when any
colision was possible was when I saved the batch file, but it really
never happened. During some of bad days, I had to make changes on the
spot, so I created two or three versions in a row, and all of them were
running in parallel :) - there was no way and no need to call 40 users
and tell them to exit and reload (called only those who complained on
the previous one), it worked well this way.
Of course, tomorrow only the latest one was in use, and I told the guys
there to delete the rest.
Now, I don't think this is easily done via icons...