john.allen wrote:Hi Morley,
I just happened to notice this post of yours of a week ago or so.
In my solution Rich Client via a server is about three to four times slower than Servoy Developer via localhost. Thus, coding which seems fine in Developer can be found unacceptable when served to users.
Based on this I would definitely say that there is some network bottleneck going on there as Robert suggested. The regular Servoy client is always faster than Developer in my experience unless the client has some network issue. This stands to reason as there is so much more going on in Developer. Sometimes, when I notice a little pause running some method on Developer, I look at it using Client (with the same network access) and it is always instantaneous. If you ever notice things are faster in Developer than the client there is definitely something going with the network setup.
Hi John,
Let's be clear we're talking about the same thing. With Developer I'm running two ways. Principal development is via Servoy's default environment using localhost, i.e. no network at all. The advantages being I don't worry about mangling my development test data and I don't disturb the production server. Performance is always almost instantaneous.
I also test with Developer pulling data from the production server via the internet. This is always slower than via localhost. For instance, the loadRecord() function can take a second or two, occasionally more, instead of a few milliseconds. This form of testing gives me a preview of performance bottlenecks without having to commit a new release to the server. Servoy Client over the net, on the other hand, is somewhat faster than Developer over the net. I've not performed an exact comparison, but that's the impression.
In my original posting I was hypothesizing the possibility Servoy3 could move towards the performance of Developer localhost. For the following reasoning. By running a virtual client locally on the server, Servoy3
may be more efficient than the dialogue of Servoy2 client to server to client. I emphasize "may" because there are several other factors at work here as well, such as the new additional task of generating HTML and then drawing the screen. So, in some ways, my speculation was also a hope.
Robert seems to think Servoy3 will be slower than Servoy2, all other things being equal. I've heard nothing to confirm or deny his expectation. I'd be very interested to know the results of some field testing.
Let me know if I was understanding your points correctly.
Kind regards,