Project Open Letter, sign it!

*bump*

You can still sign it at http://www.projectopenletter.com/

I think it was a nice try but I have yet to see a response from Phillip nor do I expect to see one. Instead we get voice, identity theft and sculpties. Some customer service...

Well LL has claimed that 70% of their effort is fixing bugs but that isn't apparent in the recent updates.

We have three new major features either in or on the horizon (sculpts, voice, clouds) but not the corresponding 7 major bugs fixed.

1.15, 1.16 haven't really solved any of the major problems (inventory loss etc.) but have in turn introduced new bugs (including the yet unfixed inability to leave group IM conversations which have no rendered group IMs unusable). 1.16 even re-introduced bugs first introduced by 1.15 and subsequently fixed!

In the case of inventory loss - the fixed announce sound as if LL are trying to fix each possible scenario when an error can occur rather than tring to build a backend architecture which is resilient to errors (which will still occur from time to time if only due to network problems).

I know that some of the new features are from third party or third party acquisitions - however given that 70% of resources are on fixing things, perhaps the next third party acquisition should be technology for resilient distributed databases rather than pretty graphics! Whilst long promised new features (havoc 2, wed rendering on prims, etc.) are just forgotten.

My suggestion to those behind Open Letter is perhaps the next thing should be a demonstration to LL of the unhappiness not just a letter. I would suggest a no-trading day. All the signatories who own shops etc. shut up their shops for a day with a poster explaining the reasons, landbaronos etc. stop buying and selling land, the general populace put off going shopping for a day etc. Basically aim to make a one day dent in the financial figures that LL so proudly display!

If that fails perhaps follow it up with a no trading weekend...

They're actually doing both. I suspect that it's somewhat of a "divide and conquer" approach, where these "bandaids" are deployed out on the live grid, tested for sanity and integrity, standardized, and rolled into the new architecture. This particular fix, having alot to do with transaction integrity, is a very significant accomplishment.

Linden dollar sales by the Supply Linden account are down 60-percent for the month of May (with a few days remaining). They were down about 20-percent comparing April to March. That's an actual dent for LL as this is a source of revenue for them. For a period of a few months this revenue was equivalent to the revenue derived from Premium account fees, though the vast bulk of LL's revenue remains tier fees and monthly island fees. In other words, money is going off the table (though I have no idea as to why).

As tier fees and monthly island fees are LL's largest source of revenue that's what you'd have to focus on. The downside is if one Premium account sheds land to reduce tier, and thus deprive LL of revenue, another resident will absorb the real estate.

The fact that it only affects only one direction (world to inventory and not inventory to world) suggests that this fix too is just a bandaid rather than adding transactional management to the database operations, but I could be wrong.

The fact that there is a report in jira of a loss of a no-copy item when taking it into the inventory subsequent to the upgrade suggests that it deals with a specific case where this happens rather than adding transactional management to all takes.

The fact that this fix has introduced a new bug whereby the object vanishes from the world and does not appear in the inventory but leaves a ghost of itself (which can't be seen or selected, can't be returned; but does take up prims; does have a physical presence as regards collisions and does still have scripts running) undermines any accomplishment somewhat.

well ll isnt the first outfit ive seen that behaved like this.
one that springs immediately to mind is madrona park/infopop/groupee - makers of ubb. anyone remember them? look where the 'we are too important to care' attitude got them. a competitor came into being, borne of users' frustration, and utterly smashed ubb into near nonexistance.

Today's blog is a good example that LL do not give the impression that 70% of their resources are dedicated to fixing problems - or even that they care.

It is quite clear from the forums that far from fixing loss of inventory on take, they have made it worse. There are far more reports of inventory loss on take since the 1.16 rollout on the forums that before. Not only that but the way this now manifests itself is that the prims are still recorded in the land prim count until the sim is reset.

What is todays blog entry - an announcment of a fix? a confirmation that LL are aware of the problem and are working on it?

No - an announcment of a new client with pretty clouds!

Today's blog is a good example of the power of open source, and the rapid pace with which something like "pretty clouds" can be implemented as a result. Don't forget that the developers of Windlight did the vast majority of the integration work - you make it sound like Linden Lab's developers have been doing nothing but working on this integration, as opposed to fixing bugs. I strongly doubt this is the case.

Zaphod,

You missed my point entirely - reread the first sentence of my post "LL do not give the *impression* that 70% of their resources are dedicated to fixing problems"

It isn't me making it sound as if "the Linden Lab's developers have been doing nothing but working on this integration, as opposed to fixing bugs", my point is that it is Linden Lab's own blogs making it sound like this.

Last week, after a weekend of shopping impacted by the llGiveInventory bug they introduced on Friday, what was the first blog entry to greet all those running businesses checking the blog to see when they could switch their affected vendors back on - a length post about fluffy clouds.

This week, after a weekend of people suffering inventory loss and land prim allownace on taking objects - another lengthy entry about fluffy clouds.

A lot of people feel that LL are adding new features at the expense of fixing existing serious problems - regarless of whether that is true or not, LL's own blog enforces that impression.

However, given that 70% of resources are in fixing the current problems - I presume the next two technology acquistions for LL will be technologies for running large scale reliable transactional distributed databases or developing and maintaining complex software projects rather than graphics.

I think that people who are getting this impression from the blog are assigning a purpose to the blog which it isn't intended to service.

If you want to know where things stand with issues/problems/bugs, then JIRA is where you need to look. That is where they are focusing efforts to communicate with Residents about issues.

Maybe we should post a new 'problem' in JIRA: "Fix all the old problems before you introduce new features!" I wonder how many people would even bother to vote for that... or bother to log in to JIRA to begin with... hehehe

*snickers* More people know the blog exists than know the JIRA does. After all, the blog headlines show up in the login screen! Maybe if they had the JIRA "Top 10 Show Stoppers" show up in the login, people would! Not to mention that this would shame LL into fixing them.. maybe?

Well, apart from the fact jira is only really suitable for the more technical savvy, and as has been pointed out unlike the blog isn't referenced from within the client, doesn't have a link from the front page at www.secondlifer.com and is general not well known about, etc...

if you go to http://www.secondlife.com/status you will find the following text:

"For additional details about system upgrades, maintenance, outages, or other known problems please read the Second Life Blog."

The blog also has an explicit category "Bugs and Fixes".

That may explain why people think that the blogs purpose is to inform us about known issues, bugs and fixes!

There already is one, I posted it earlier in this thread.
Although I do struggle getting JIRA to let me log in sometimes

All well and true. Sorta. There are *hundreds* of issues being tracked in JIRA. The issues that tend to make it to the blog are issues that they feel need to be addressed to the widest possible audience.. they are issues that often don't ever go into JIRA, as they are relatively short lived "break/fix" issues. The password reset issue recently blogged for example. Otherwise, they blog on scheduled maintanence, upgrades, planned or unplanned outages, etc. The meat and potatoes of issue tracking is in the JIRA. No, it isn't a very friendly environment, I'll give you that.

So what do you want to see on the blog, given that there are currently 160 "Service" issues, and 544 "Viewer" issues being tracked? In total, there are well over a thousand issues. None of them are resolved, many of them won't likely be any time soon. So what sort of blog post are you expecting? Bird's eye view/roadmap? Read Cory's townhall transcript (both of them). Status on specific issues? Read the JIRA.

Zaphod,

Just the problem I mentioned in my original post - namely that the fix for inventory loss on take in 1.16, now intermittently results in

a) the item not appearing in the inventory (i.e. is lost if the item is no-copy or a recently created item)
b) the item although not "present" in the world still interacting physically (via collisions) and still using up the land prim count
c) any scripts in the object continuing to run
d) as the object is not really there, it cannot be selected, returned - even an estate owner can't return it or shut down the running scripts.

This new problem has been generating multiple threads on the forums per day since the 1.16 rollout - if the number of threads in the forums is in anyway indicative of the frequency this occurs, this is happening far, far more often than the inventory problem 1.16 was supposed to solve.

When it does occur it is causing loss of items and work, reducing peoples available prim counts, rogue scripts which cannot be disabled even by the owner or estate owner etc.

So far there is no acknowledgment of this new problem or when it may be resolved from LL.

I found two JIRA entries on this: SVC-242, which has exactly 1 vote, and MISC-237, which has 3. SVC-242 was entered by Which Linden, so there's your acknowledgment of the problem. The forums are generally not a great indicator of how wide spread an issue really is, considering we're now getting close to averaging 40k concurrency. The issue has been acknowledged by Linden Lab, and if it is a hot problem, it needs to be escalated in JIRA, not blogged about on the blog.

As regards the Jira entries - the original entry from Linden downplayed this as somethiing to note! There is no indication that LL are working on fixing this or just lumping it with all the other inventory issues. Actually LL feedback on progress on any issue within Jira is often quite poor, even within Jira, with quite major issues being marked unassigned unless you prod LL hard. Part of the problem is that LL seem to have their own internal bug tracking tool - Jira works best when it is the only bug tracking tool and reporting mecahnism, as is really designed for use by (typically opensource) development teams

More of those 40K concurrent users know about the forums (and post here when problems happen) than Jira (there are plenty of examples in the forums of people not knowing what Jira is, and I am yet to meet anyone in world who knows about jira).

Of those who know about Jira don't use it - it is complicated to navigate, can be slow or even down at times, logging on from anywhere but the front pages doesn't seem to work in some browseers (e.g. Opera), and as mentioned feedback from LL in it is poor.

People do know about the bug reporting tool built into the SL client though - however that doesn't link up with Jira. In particular, reporting the problem via that does not bump up the number of votes on the corresponding Jira entry!

Hence the vote in Jira not a good indicator of anything, and certainly no better (if not worse) that gauging trends in the forums.

Anyway, 1.16 explicitly listed Inventory loss on take as fixed, subsequent to 1.16 there has been a far larger number of threads in the forums of inventory loss on take than before 1.16, but now with additional side effects (prim count on parcel not reducing; scripts still running etc.). The fix in 1.16 has made the problem worse not better.

Given this is a thread on the OpenLetter - updates on the other bullet points raised in that might also be appropriate.