Warning: Emerald and Shapes

Oooh, a stab at the Jira?

Jira is without doubt the best bug tracking system I have ever used. Bugzilla is one of the better bug trackers, and way ahead of almost every commercial tracker I've ever used... and it's nowhere near the quality of Jira. There is no better tool Linden Lab could have picked, as far as I can see.

As for Emerald, it's an experimental viewer. If you're not comfortable running an experimental viewer, where every release is at best RC quality, you shouldn't be running Emerald. Stick to the standard viewer or one of the forks that are concentrating on stability instead of features.

heh, the kool kids rally the wagons.

Phil has good points. Also, don't expect regular users to go out of their way to report bugs - there should be no expectation of this at all.

But like Argent said, its experimental viewer so i guess that says it all.

I just don't like to see the kool kids moaning at people complaining.

Phil.. I exported my shape, my skin and a scripted on/off firebowl with sparks from SL and imported it to OSGrid.. worked fine.. I was very tickled. I was creator on all stuff including scripts. The objects do have to be rezzed for that to work so I put my shape in a box and exported it.. then opened box after import, the firebowl was just exported as is. I need to figure out how to set up a sim before I piddle with it much more. I'm using 905.

On the shape thing, I always Save As on my shapes because I'm always tweaking my avatar, so it's easy to step back one if it breaks, although I jiggle and have edited many times and had no problems. I give kudos for the Emerald people telling us quickly when they find bugs.

It wasn't a stab at the Jira, Argent. It was a statement to the effect that, since normal users of Emerald are not faced with suggestions to report bugs (don't use the Greenlife website), there's no reason why they would. A few would, of course, but most wouldn't. It was part of my reply to Kitty when she asked me if I'd reported the bug and worked with a developer on it. There's no suggestion in the viewer to do that and, if it's not a serious bug, most of us wouldn't go searching a website looking for a way to tell them about it.

Is there anywhere that says it's an experimental viewer? I assume that it packs more added features than at least most of the others, but that just means it's better than most of the others I'm still in love with the temporary uploads (not really uploads) that allow me to try textures, animations, and sculpties on the main grid, and without cost. I'm using that feature quite extensively. I like other things about it too - double click to TP to somewhere you can see, TP to a person in the radar box, and so on. I'm not against Emerald at all - I'm fully in favour of it. It's just that I think they are putting many new features in it, too quickly - without extensively testing them all. Remember that most users of it are probably just regular SL users who don't ever think about these things, and probably still aren't even aware of the shape bug.

I'm not a bug reporter I've only ever reported 2 in SL. One was a bug in the mono implementation on the beta grid - the MLP system wouldn't fully compile to mono. It was fixed before it was released in the main grid. The other was in the Classifieds part of the profile in an RC. That was fixed before general release too. But normally, I'm like most people - we think that they must already know about because someone will have told them

I do think that Greenlife would do well to institute a Release Candidate system. Then those who have the nature to report bugs and such will do so more readily, because they'll intentionally be part of the testing of new features.

Please don't twist my words into something I never said.

You're pointing a finger and making assumptions that things aren't tested just because there happens to be a bug in something. It happens, and my point was that whether or not it was tested thoroughly is not something *you* can know.

As for RLVa which I'm solely responsible I can say that plenty of testing takes place: be it tracing over code line by line to make sure it's doing what it should be doing and that nothing unexpected can happen or unit testing or trying out various scripted toys.

In spite of that bugs still turn up: in some cases I can indeed say that I should have found that myself because it's "obvious" but in other cases no amount of testing would have turned up the bug because it relied on a specific set of circumstances (ie getting disconnected from a sim while under a specific set of restrictions would cause a crash because there's no longer an active region).

The only thing you can do in that case is fix the bug and go look for other similar circumstances and be mindful of it in the future (and for the latter example it turned out that there's a handful of things where the official viewer will crash under the same set of circumstances).

---

Or a bug in the official viewer that used to drive me crazy: http://jira.secondlife.com/browse/VWR-14765 . I could do a rant that whichever Linden added the grid selection combobox simply didn't test the silly thing since it leads to a crash if you do 1-3! Or how they should have realized it while coding it already since it's so obvious that clearing the pointer in A is going to lead to a crash in B!

Except of course that it's not nearly so obvious - except in retrospect - and you could spend a month testing that little bit of code but unless you randomly hit the magic combination listed in the repro you're not going to run into the problem.

---

"Bug" equaling "not enough testing" may or may not be true, but it's not something you can know.

Users definitely shouldn't be used as guinea pigs to vet code as a matter of habit, but given the impossibly large set of possible circumstances that can all come up they're guaranteed to run into situations you didn't and in those cases the only way to know there is even a bug is because it gets reported (even more important for platform/hardware specific bugs).

As far as not wanting to report a bug: that's fine, you don't have to. Be it in the official viewer, or any other. Your customers also don't have to tell you about a problem they run into with something they bought from you either, but if someone came on the forums complaining that you're selling broken/half-finished stuff without ever bothering to contact you then you'd probably argue that they should have tried contacting you first to let you know about the problem.

Oh, good, Kitty, you're an Emerald developer? I've been all over Modular's website, and the only way I see to report bugs is to send info to a developer. Is there a public reporting mechanism, or even a template to use?

I looked at the Google Code site, and the issues list there is obviously not in use. Is there a Bugzilla instance somewhere for Emerald?

Even more than *reporting* bugs, I would love to *see* a current issues list. Is there one being maintained that I missed?

(p.s. I asked this in the inworld group, but the question was ignored)

If they don't use the greenlife website where did they download it from?

The rate at which they add new features should be a red flag to any old hand at SL, but you're right... they're not making that nearly clear enough on the website. THAT I'll fault them for.

Only in the broadest sense of the word I guess . I integrate RLVa into Emerald but that's really all of it as well since it keeps me busy as it is (there's a few of my JIRA patches in it as well I think).

There's the "bug report" forum on the site ( http://modularsystems.sl/index.php?option=com_agora&task=forum&id=3&Itemid=2 ).

There used to be an actual issue tracker as well, but I think it's fallen into disuse . You can contact LordGreg in-world as well (or on Skype) and he will follow up on it as well (or at least he passes RLVa related issues on to me in any case so I'd think he does the same for other problems ).

Not that I know of personally but I'll try and nudge for one somewhere .

I wasn't the only to understand that it's just down to the developer when you wrote:-

I didn't twist your words. I may have misunderstood them because they weren't clear enough for what you meant, but I didn't twist them. Having said that, lower down you confirmed that I hadn't misunderstood them.

You're right that I can't "know" how much testing has been done. I was going by what you wrote, and by the fact that we've had a little flurry of things not working quite right.

I don't know what RLVa is, but what you wrote there does suggest to me that insufficient testing has taken place. One person (you) going over lines of code and no doubt trying things out in SL, isn't extensive testing. Even LL don't attempt that with viewers. After they've done all that, they then release as an RC for extensive testing in real use, before they release it as the normal viewer.

Agreed. And I don't think that I've ever stated that not enough testing takes place. I always qualify such things with things like "it seems to me" or adding a question mark - e.g. "too much too quickly?" I also read reports of bugs in this forum and put 2 and 2 together. And from you wrote (quoted above), it does sound like I add them up to 4.

I don't sell broken or half-finished things, Kitty, so that analogy doesn't apply. If I did sell such things, then I could rightly be criticised for selling things that haven't been fully tested.

Don't get the idea that I am against Emerald in any way. The opposite is true. There is no need to get all defensive about it. In two posts now, you've indicated that extensive testing does not take place and I happen to think it should - as far as is possible, anyway. It does seem to me that Greenlife are packing Emerald with too many new features too quickly and, as a result, some of the things are bugged, or simply don't work, probably due to insufficient testing. Pleas note: I said that "It does seem to me".

There's a big difference between using a website in the way that I obviously meant, and getting a download URL from someone and downloading a piece of software. For instance, I download software updates from time to time. They come from websites, but I'm not a user of those sites, other than the minute it takes to spot the download button and click it

You're asking for the fuckup fairy to drop something nasty on your machine, you know.

Quite possibly, but she hasn't visited me yet

Kitty:

I am a fully paid-up supporter of Emerald. It does wonderful things for me. Don't get the idea that I am against it. I even started a thread a week ago to sing its praises, and I still sing them. I noticed something, so I mentioned it. That's all. I also suggested what I am sure would be beneficial - an RC system. That was constructive. There is no need to be on the defensive.

Emerald does not get near enough "polish" as it should. It is in all senses a moving release candidate, and experimental at best. To say that there is an actual team behind Emerald would be a joke. While we do work together at times, it is mostly individual effort for any given feature. Some features have large missing gaps in the code and are unfinished(import/export), some have bugs that we did discover until things were put out.

People would be right if they did not treat Emerald with the same respect as the official client in regards to how it is released. It does not have an extensive bug testing period and is often released "as is". Saying it goes out entirely untested is unfair as well. We do take some steps to ensure that blatant and large bugs that we are aware of do not make it to when we "release".

I'm happy to see it has gained the overall reception that it has, but its a victim to its own success as well, we spend a large amount of our time doing public relations instead of time we could be doing development.

I don't think anyone said that "it goes out entirely untested", Zwagoth. I've said that it seems like features go out without enough testing, and your post appears to confirm that. And I'm a fan of Emerald becuase of some of it's features.

It's not just a case of not enough testing, it's more of a case of not the right kind of testing. Developers by their very nature usually only test the component they've built, sometimes if you're lucky all the components of a function are tested together. What developers never do is test it in a fully application integrated form. i.e they may test the function in isolation but they may not test it in a "production-like" environment nor with any kind of real life user workflow.

an example of this is here: http://news.bbc.co.uk/1/hi/technology/8328399.stm

the devs used linux which didn't pick up the scareware.

There's never enough time to test every single facet with every possible option (even if you could think of them) which is why risk based testing is a more sensible option.

I do second Phil in his suggestion that you look to a rc issue prior to formal release. there should be a critical mass now that you should be able to pick up a few people to run through any changes and give it some end-to-end functionality testing before it goes live. It still won't overcome all of the problems and from the errors I've seen that people are most concerned about it may not even have found them but you just never know until you give it a whirl.

Feel free to start an Emerald QA team.

QA isn't testing. Which is another mistake a lot of organisations make and then wonder why their deliverables are so crap.

For years I wanted to get into LLs IT teams and give them some structures that would improve the quality of their releases. Much as I would like to show them how it should be done (at least within the test arena) all it would probably do is drive me crazy, my way of working is too structured for most in here.

To pull it off with any degree of confidence I'd need a group of people who understand the concept of collaboration and yet are disciplined enough to be able to follow instructions rather than just having a play (as you know, there's a place for that but not to the exclusion of everything else) or doing what they think is best, because of the frequent releases for emerald they'd have to be available within timeframes and the devs would have to provide documentation on what they'd done so the testing could be targetted appropriately.

now that is me lowering my standards to the bare minimum I can cope with before I start thinking I'm losing hours of my life that I'll never get back.

Getting people to work as a group in here is like herding cats at the best of times, let alone the rest of it. Add to this the fact that I drive most people crazy and it's just not viable.

I have a lot of respect for the emerald team for delivering such an impressive package. Distributed dev teams are not optimal for delivering quality yet Emerald have managed it and outclass LL by a factor of many. It would be good to give them the final piece of the lifecycle but I'm not the person to do it.

Well, they don't have a bug list, so they are far far from thinking about "lifecycle"!

We are Emerald's testers, and that's not going to change. So I concur with Phil and others who have suggested a release candidate system - it's easy to do, and would act as a flag for those people who want a more stable experience. For example, at this moment I think the "released" Emerald might be the .673 build, and both 904 and 950 (I'm doing build numbers from memory here, forgive errors) would be release candidates that would never go to release.

I think if they did that and made some attempt at providing real release notes (for releases, not release candidates) with a "known bugs" section, and a web page where people could report bugs for each version without having to join Modular's forums, they would be in great shape.

they may not have a formalised jira (thank god) but they do monitor the various fora and they do take notes. It's formalised adhoc if you like. Internally I don't know what they have - they could have a traditional method of xsl spreadsheets or it could just be emails with the details to the developer responsible for that area. This is an organic development of their methodology and these people aren't fresh off the boat when it comes to IT development so whatever they have will be what works best for them within their current constraints.



If you've got people reporting bugs you need some kind of system to be able to get in contact if you need more info. Anything they did with regards to your wishlist would have to be on their website and you would need to register to reduce the incidence of griefing/trolling. Emerald already gives then your account details so having to join the forum is really not much more of a hardship. What would be good is an integrated bug reporting system in the browser - that way the logon to the website would be automated and would make the capturing of defects easier. That's a solid piece of work though and without knowing the chance of people using it may be a waste of effort better spent on other things.

More information would be a very good start though. Release notes are important - particularly when sometimes it's the only place the changes are actually documented.

Account details are not stored/kept by anyone to do with Emerald, and neither should they be. Some people have a lot of money in their accounts and access to it should never be given to anyone other than LL. So any website login would need to be completely seperate and should never ask people to use their SL password.

that was sloppy thinking and typing on my part. The first point i was trying to make that was for those who are resistant to signing up to websites is that if you use emerald then you already trust them to manage far more important details (i.e. longon details) than you'd be giving to the site.

The second point is that if the request to the webpage is coming out of the client then certain validations have already been met (I did forget about spoofing though) but bringing up a page where all that needs to be entered is a website password is far more user friendly than expecting someone to search and find the page then log in and then doing the reporting.

Understood. I just wanted to point it out in case anyone got the wrong idea about Emerald and their own account details.

sensible. There are a lot of misconceptions about emerald that don't need to be fanned.

No, but testing is a crucial element of QA. In any case, if you want higher quality in Emerald, you're free to apply your knowhow; I'm sure they'd appreciate your efforts.

It's a volunteer thing. If you want something done, you can ask and hope, or you can take the initiative and make it happen. Your choice.