Question: Is there a parallel universe? (or just a parallel EU loader?)

[Naomi]

If I were head programming for MA, I would not work with a parallel universe, but with a parallel loader. I would implement it by using 2 loaders, i.e. a 'client loader4 and a 'developer loader'.
The first giving access for the gaming community, and the second giving access to all possible stuff for developers, programmers, MA staff etc, including an area that is not available for public, where MA can test some stuff.

I think this is a very logical approach of EU programming, and this is possibly the way it works. This has the following consequences:
- New features are first only accessible by login in with the developer loader, which means they can be accessed from before the official launch. This means that all new features are already present before launch (but not downloaded by the clients yet).
- In case of a VU update, the file list that is used when starting from the client loader must be updated, so that new content, files become available to the public.
==> Client loader and developer loader both have a list of files that can be accessed/area's that can be used etc.

So, if all works this way, I can't keep myself from wondering:
WHY IS A LONG DOWNTIME NEEDED? ISN'T it ONLY A MATTER OF ADAPTING THE FILE LIST OF THE CLIENT LOADER?;
Even more: with this system, I think it must be possible to adapt changes the hot-plug-way.

cheers
Naomi


<disclaimer> This is in no way a negative post. It is only a suggestion how I think things work, or could work. Maybe there is room for improvement, or maybe I'm just too noob in understanding how it all works? Feel free to discuss (not my programming-noobness, but how it works :d)

 

[girtsn]

I have said it before and it might be my delusions.
Have noticed more lags than usual before VUs, which has led me to believe MA is using same infrastructure to test/stage as the one running production.
Don't like the effect this has on performance, however it's a fine approach for QA. Unfortunately it is not working too well anyway (I understand the dependencies are huge but some bugs are just... poor QA and inability to properly fix)

I think update downtimes are actually needed for technical reasons, some tasks can not be performed running. E.g. database schema changes etc.