Jump to content

Deploys for the week of 2012-03-05 ROLLBACK

Oskar Linden

By Oskar Linden,
in Second Life Server

 Share
You are about to reply to a thread that has been inactive for 4427 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Ayesha Askham

Ayesha Askham

Posted
Posted

@Oskar

I am not sure how significant this is, but since the restarts I have noticed that although my inventory is set to open when I receive a new item, (Notecard, landmark or object) and accept it, the inventory stays closed even though the item is received, and in the correct folder.

So I guess it is not expected behaviour, but it is hardly critical, from my viewpoint.  I run Firestorm 3.2.2 on Windows 7 64 bit FYI.

Would this behaviour be related to the inventory handling changes in the software update?

Sim is 21071 LeTigre Woods of Heaven.

Link to comment
Share on other sites
Theresa Tennyson

Theresa Tennyson

Posted
Posted

Ayesha Askham wrote:

@Oskar

I am not sure how significant this is, but since the restarts I have noticed that although my inventory is set to open when I receive a new item, (Notecard, landmark or object) and accept it, the inventory stays closed even though the item is received, and in the correct folder.

So I guess it is not expected behaviour, but it is hardly critical, from my viewpoint.  I run Firestorm 3.2.2 on Windows 7 64 bit FYI.

Would this behaviour be related to the inventory handling changes in the software update?

Sim is 21071 LeTigre Woods of Heaven.

I went to NCI Kuula (a Le Tigre sim) just to test things and I clicked on some notecard givers. I got the notecard fine but they didn't automatically open the way they usually seem to do. Viewer 3.2.8 for me.

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

@MB and others:

It may well be an issue for Firestorm Release, but it only started since the restarts on LeTigre for me and I know others are seeing the phenomenon as well on LL viewers, so I'm covering all bases, by raising the issue here and on the Phoenix/Firestorm JIRA.

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

OK an update:

The inventory behaviour is consistently NOT AS expected on various LeTigre sims.

Inventory behaviour is consistently AS expected on Magnum and Main Server sims, i.e. the inventory opens as set in preferences and the received item is in its correct folder.

To summarize ( geez I wish it WAS summer...) although items are received in their correct folders on LeTigre, the inventory does NOT open as set in preferences, this effect only having been seen since the rolls on Thursday.

I have not yet got to a Bluesteel sim to check it, but since it is running the same software as LeTigre that shouldn't matter.

Link to comment
Share on other sites
Lydia Craig

Lydia Craig

Posted
Posted

What improvements Oskar?  Since the last restart on the RC channel things have gotten worse and not better, logins are now taking hours due to inventory problems, crashes are more frequent, the memory leak which LL apparently chooses to ignore or cannot fix is getting larger, lag is worse, crashes are coming much faster now, other loging problems like suddent unexplained crashes have begun and in general service is as bad as it was at its low point in December.  It would be nice if LL stopped its constant barrage of new and untested or poorly tested programs and made fixing these problem a priority, and actually followed through on that committment, rather then giving lip service to it as they are now.

People are literally paying hundreds of dollars a year to play SL and are recieving absolutely pitiful service in return and all to often ar placed on RC servers without being asked and submitted to what amounts to a paid beta for service that would be shameful even on in alpha testing.  In light of this is it no wonder that online numbers are down, old players are leaving sl in droves and new players are giving up in frustration at an ever increasing rate.  LL has reached the point that if really wants to save SL and not destroy it, as it is doing now, it needs to address these issue, establish some degree of stability and decent service and do it quickly before there is no one left but bots to appreciate the effort.

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

@Lydia

You obviously have a problem, Lydia, but the memory leakage is client side, not server side...try a different viewer, that might help your stability issues.

LL have certainly got issues to address, but these posts are becoming monotonous and are clearly being ignored, so save your bloodpressure and lighten up!

Link to comment
Share on other sites
Sassy Romano

Sassy Romano

Posted
Posted

Oskar Linden wrote:

Second Life RC BlueSteel
  • This is an InventoryAPI maintenance project.
    • Improvements were made to the way inventory transactions are handled in order to improve scalability and reliability of inventory transfers.
       

Second Life RC LeTigre
The code in this channel is identical to the code in the BlueSteel RC channel.

This inventory project and the way that it handles inventory has at present, broken all products that are RLV and send inventory offers which would have othewise gone to #RLV.

I appreciate that RLV is niche, not of interest and not in an LL viewer but nonetheless this has a massive impact on RLV developers and the support that is needed unless this can be addressed in the viewer.

Even so, there's a transition period and the support in the interim will be huge.

Yet another episode in the recent "lets break stuff" story.

https://jira.secondlife.com/browse/SVC-7748

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

@Oskar or any Linden Lab employee that works at weekends

 

With llgiveinventory badly foobarred on LeTigre and Bluesteel RCs: ( JIRAs SVC-7740 proven and acknowledged, SVC-7745 so far NOT acknowledged), I see Sassy points out another point of failure of this code change.

Since the Phoenix/Firestorm team rapidly triaged this issue and pointed me to these JIRAs, it would not have been unreasonable to expect some comment on this thread from The Lab.  But wait...the Lab went home on Friday at "end of working day PDT". So, no reply.  "Aura Linden" must be weekend cover.

You know Lydia bangs on about poor testing, and some of us think she goes over the top, but this is a classic example of what she means.  How on earth could something like this be missed in QA??

Answer: 'cos it wasn't tested inworld. If it was, the tester was incompetent, end of story.

This isn't snark, Oskar, this is us giving LL's QA a bloody nose.

Edit: Oops. Thanks to Qie for pointing out my numerical dyslexia...JIRA nos corrected, and Qie posted the links properly.

Edit 2: SVC-7748 for RLV issues.  This one is a gamebreaker for many.

Link to comment
Share on other sites
Theresa Tennyson

Theresa Tennyson

Posted
Posted

Ayesha Askham wrote:

@You know Lydia bangs on about poor testing, and some of us think she goes over the top, but this is a classic example of what she means.  How on earth could something like this be missed in QA??

Answer: 'cos it wasn't tested inworld. If it was, the tester was incompetent, end of story.

 Yeah, they should have put things up on Aditi and given people a couple of weeks to test things there.

 

OH WAIT - THEY DID

 

Link to comment
Share on other sites
Theresa Tennyson

Theresa Tennyson

Posted
Posted

Ayesha Askham wrote:

@Theresa

So, you say that:

1) we had an ample opportunity to test it?

or

2) The tester didn't notice glaring inventory errors when testing an inventory API?

 

I'm unclear on your stance here.

I don't know exactly what code was put up in the test regions but they were up for at least a couple of weeks and they asked for people to come in and test it. It's really difficult to catch an error in something you do yourself or are very familiar with. It seemed to me that that more people crossed their arms and said, "No sir, I don't like it" to the general concept than actually went in and experimented.

Regarding the errors: The RLV problem is serious but it's with something that isn't part of the standard code and it sounds like the methods RLV uses to send things to specific folders go pretty deep into the nuts and bolts of things. It would be hard to find if you're not familair with RLV. I can see the problem with notecards, etc. not opening automatically but that doesn't strike me as "glaring." And the problem with changing outfits flat-out doesn't happen for me where it does for others and I've been TRYING to make it occur.

There are things that need to be fixed in this release. That's why it was sent to the RC channels.

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

Theresa

I have two points to answer here.

1) the inventory errors ARE glaring.  No, they in themselves are not serious, but they indicate a serious failure in the code.

2) I did not at any time give Linden Lab permission (nor was I ever asked) to use me and my sim as a beta test area.

Such errors should be worked out of code long before it reaches a Release Candidate status.

The RLV issue is due to a serious error in understanding how inventory works on the part of the developer, and if you think that RLV is not mainstream, you are badly mistaken.

If Linden Lab are trying to alienate users, they are doing a fine job.

SVC-7745 does seem less reproducible, so I'm not sure what that indicates.

I may have this wrong, but you appear to be saying that because we users on the Main Grid didn't do Linden Lab's work for them on Aditi, it's our fault...eh?

Link to comment
Share on other sites
Theresa Tennyson

Theresa Tennyson

Posted
Posted

Okay, here's the timeline as far as I can determine things:

1) Oskar announces testing, puts up test regions on Aditi asks for JIRA with problems.

2) RLV developers test things on Aditi, discover that things break.

3) RLV developers do not file JIRA, instead comment in forum thread that's many pages of non-technical comments.

4) Code gets promoted to RC channels.

5) Stuff breaks.

6) RLV developers file JIRA.

Link to comment
Share on other sites
Sassy Romano

Sassy Romano

Posted
Posted

Theresa Tennyson wrote:

3) RLV developers do not file JIRA, instead comment in forum thread that's many pages of non-technical comments.

Yes, in the thread that was posted as the one where comments were requested to be submitted by both the commerce team AND Oskar.

One would therefore also reasonable expect that LL reads said comments.

In fairness, LL partially did but failed to comprehend their user base.

Link to comment
Share on other sites
Ayesha Askham

Ayesha Askham

Posted
Posted

@MB

I DID file a JIRA as it happens, and I did post on the server blog where Oskar says they watch closely.  I also posted on the Firestorm Support JIRA because it was suggested by someone that it might be a Firestorm bug.  I think you know who that was, MB.

And, Yes, I do believe that someone at LL didn't do their job properly.

Granted I suppose I could also spend time testing on Aditi (if I actually HAD the time) and then LL would have had even longer to ignore my comments.  After  all, they ignored the germane comments of someone far wiser than I, 

https://lists.secondlife.com/pipermail/opensource-dev/2012-February/008641.html

so why would they listen to me?

Time after time LL tell us that they are a competent commercial company. I have to say that the facts do not support that assertion.  If this mess is promoted then I will eat hay with a donkey.

I found this snippet: http://community.secondlife.com/t5/Featured-News/Advanced-Experience-Tools-Preview/ba-p/1420633 this wouldn't have anything to do with the destruction of #RLV forced wear would it? Hmmmmm.

Just because I'm paranoid doesn't mean they aren't out to get me, now does it?:smileytongue:

 

Link to comment
Share on other sites
Oskar Linden

Oskar Linden

Posted
  • Author
Posted

The issues found in this RC were always considered bugs and there wasn't an intention to promote the code. We put things into RC to find issues like this. We also beta test them on ADITI. This code was on ADITI and there was an open beta test that I encouraged people to attend. We recognized these issues by Friday morning and have been working on fixes since then. Wednesday we will update the code with the bug fixes.

__Oskar

Link to comment
Share on other sites
Sassy Romano

Sassy Romano

Posted
Posted

Oskar, feedback was solicited by LL in a forum thread for merchants.

 

I know that I posted this RLV behaviour after Kitty and I tested it on Aditi.

 

However, this still went forward to RC on the production grid so the question is what went wrong with the feedback process and more importantly how should this be corrected in future?

 

If feedback on a forum thread is what's asked for, that's where I will reply but it appears to have failed in this instance.

Link to comment
Share on other sites
You are about to reply to a thread that has been inactive for 4427 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share
Go to topic listing
×
×
  • Create New...