Given the nature of how Second Life works, and given the nature of anything net-based in general, I didn't bat an eyelid back on Friday when I tried to open a script for editing and it just didn't open. I had other things I needed to get on with so I put it down to it being "one of those things" and carried on.
I started to worry a little more when, later in the day, the same thing happened. I'd try and open the script (which was in an object's inventory) and the editor would just sit there, saying it was loading, and nothing more would happen.
A little more playing showed that smaller scripts where just fine, slightly bigger ones appeared to show an unusual delay but would then load, and then anything "non-trivial" in size (for values of "non-trivial" I've yet really figure out) would just fail to load.
I've seen this sort of thing before, of course. Normally TPing away and back, or a relog, fixes the issue. So I tried both of those things and it made no difference. Next up was to try a region restart. Again, I've seen similar effects in the past that a restart has cured so, seeing as nobody was around at the time, I kicked off the restart.
This didn't fix it either. The region came back, I TPd back in, same problem.
Next I TPd over to a plot I have on a main channel region. I rezzed out a copy of the object that was having the problem and... the script opened in the editor just fine!
I then TPd home once more, tried the same, and had the same problem.
As luck would have it a region next to my mainland plot is a LeTigre region. So, this time, testing with a script in my inventory (rather than in an object's inventory), which was failing to load into the editor back on my home region, I headed back to my mainland plot and tried again. Sure enough, it loaded just fine on the main channel plot. I then closed the editor, walked over into the adjoining LeTigre region and tried to edit it again. It failed. This time, for the first time, it failed with the following error:
Second Life: Unable to load the script. Please try again
That seemed to narrow it down. It looked like it wasn't (obviously) an issue with my setup, it was something to do with LeTigre regions (and possibly other RC regions? I've not been able to test just yet).
By this point I was running out of time and wouldn't be in-world for a day or two so, on the off-chance that it might be useful to someone else, I mentioned it in the forum thread for that week's deploy.
Today I've had the chance to do some more testing. So far I'd been doing all of this with the latest version of Firestorm so, to eliminate any issues with that, I installed the latest Lab viewer and did all of my tests again, with a different avatar too. This time I managed to make a script that was simple, could be shared with anyone, and displayed the same problem. You can find a copy of it over here. I can reliably create a script in inventory, open it in the editor, paste that into it, close the editor, try and reopen it and it fails to open.
Since then I've also done things like reboot my router, turn off any form of DoS/scan protection (just in case), etc... Nothing's made an obvious difference.
So, right now, for me at least, LeTigre regions are stopping me from getting at the source code for any of my non-trivial scripts. Thankfully I've got them all backed up locally, of course. And, for now, it seems I can go to a main channel region to edit them. That does mean though that the region I rent, the place where my workshop is, isn't much good as a place to script.
I've now filed a Jira for this so, hopefully, one way or another (an issue with my setup, or a server issue), this'll get sorted in the next week or so. If not, I'm sort of worried about what this means for existing scripts that, until last week, I could get at.
If anyone else can reproduce the above I'd love to hear about it.
Update 1: I've had one person report back that they are having no trouble loading the test script on LeTigre regions, including the one next to my mainland plot. So it's not a general issue then. Damn. :S
Update 2: I've now tested this issue on the Main Channel, BlueSteel, Magnum and LeTigre sandboxes. The script loads just fine on all of them apart from LeTigre.
Update 3: I've tested this on a different machine on a different network and with a different Internet connection and, there, the script opens just fine. In fact, it opens faster than I'm used to most scripts opening. So there's something particular about my normal environment and LeTigre regions.
Update 4: For the sake of completeness I've tested on a different machine, on the same network and Internet connection as the machine I was initially having the problem with, and I get the exact same results there (won't load on LeTigre, loads elsewhere).
Update 5: Interestingly, the same test script, pasted as text into a notecard, loads just fine.
Update 6: The problem has been identified and a workaround to let me work on big scripts has been found too. I've just had a bit of a chat with Andrew Linden, who did some tests while I reproduced the problem, and the cause has been identified. He also suggested a workaround that works every time for me. If I lower my viewer bandwidth to around 200kbps (normally I'm on 500 and, while trying to get to the bottom of this myself, I'd actually upped it to 1100kpbs using this formula). The moment I dropped my bandwidth to 200kbps the large script started loading again without any problems. Yeah, I'm a bit relieved. ;)
Update 7: After the 2013-06-05 rolling restart everything is back as it was. I can now load large scripts again as I used to be able to. The code that caused the problem has been backed out, from what I've been told. What is exciting is that I tested some new code over on the beta grid the other day that had large scripts loading for me in less than a second. "Normal" loading of large scripts for me right now is around 5 to 8 seconds. Hopefully that change will hit the RC channels in the near future.
No comments:
Post a Comment