It is a handy guide, which I followed through to work out how to build from source after I found all the binaries I downloaded crashed on start and wanted to tell Microsoft how unhappy they were.
It was fine up until the configuration of opensim.ini, where it is no longer up to date with the current changes (post 99xx) where the INI files split up into main and grid/standalones. I went back to an earlier release which conformed to the guide and managed to get a standalone running exactly as they described. But I was lost with the 9952 onwards releases, despite other people on the forums posting thier solutions.
I persevered with the ini files, and still was not able to get a standalone version later then 0.5.1 post-fixes running, even after switching from SQLlite to Mysql.
Finally, I was told how to work round the loopback issue with client and server on two separate machines on the same LAN without having to get a DYNDNS account or faf around port-forwarding on my router.
I then worked backwards, having got my system running on MySQL, to find that even with the loopback workaround, SQLlite will not work in the 9952 and upwards issues.
Useful though that guide might have been, the fact that it was out of date with respect to recent changes made it no real help at all for me after managing to build the releases.
All in all, I spent an entire week trying to get the latest (0.6.6) version running on a standalone. I could have stuck with 5.1 post-fix, of course, and waited for somebody else to sort out how to get 6.6 running. I wonder how many other people have come along to have a look at Opensim and drifted off in confusion?
After the discussions in LBSA plaza this morning, I understand that a user-guide is to be produced to help both newcomers and existing users get up and running with the newer releases. Good news, but why where the problems not detected earlier, at the pre-release stage?
Don't take my post the wrong way, I am not knocking OpenSim or OSGrid at all, just pointing out that your main inrush of newcomers that you want to make all those green dots start showing up on the map are not likely to be willing to lose a week of their lives to log in and find themselves Ruth on a desert island.
I think that a useful pre-release test for any major version coming out would be for somebody to take it and build a standalone sim from cold, (no previous MySQL database), to ensure that a new user can start from either binaries or source, no need to download MySQL or go out and buy loopback-enabled routers, and get to Ruth standing on a dome of land in a rolling blue ocean in a reasonable space of time. If problems are discovered, such as SQLite not working, then at the very least a readme could be packaged with the release to tell people they need MySQL. Otherwise, I see lots of potential OSG users just walking away and waiting until somebody sells them an out-of-the-box solution, or just going back to Linden.
A subsequent test, looking at some of the other posts, is to make sure that the newly-built standalone upgrades an existing standalone without losing all the existing data.
I am happy to volunteer to do pre-release testing on Windows XP 32-bit to see if a release candidate can create a fresh standalone, but I am not sure exactly who I should approach to offer this service to. Answers on a notecard, please
