I have used a loader for many years, and I love it. Whenever I try a newer version, I simply run it on a workstation. After it's tested and accepted, I copy it to the server and send a message to all the users that the new version will be active as soon as they exit and re-enter their program. If someone discovers a blunder, and I must admit that it has happened, I change the date of the last working version to date(), and copy that to the server immediately. Works like a charm!

>Yes thats what I was told.
>
>So a loader should also check for the latest version and move that to the local PC and then run that executable.
>
>Though I only have various opinions to refer to and I don't know the full story of each perons individual network configurations.
>
>
>
>>Hi Hilmar,
>>
>>I remember the discussion, where someone mentioned that they had performed many tests, and the conclusion was that the exe in many circumstances could be transferred from the server many times in small chunks. So running the exe from the local machine can be a big boost.
>>
>>>>Hi Hilmar
>>>>
>>>>I had a conversation with people here a few days ago which seemd to indicate that running the exe directly off the server was a performance hit.
>>>>
>>>>You seem to not find this a problem which is interesting.
>>>
>>>I suspect it would delay the startup only, although others thought that it might make the performance after startup slower, too. The real performance problems - IMO - come from transferring the data over the network.
>>>
>>>Although I didn't seem to have serious performance problems, this certainly is an issue that warrants further investigation.