Weekly updates = bad development idea

Lindens -- I don't know how long your actual development and QA cycles are, but with weekly S/W deliveries I can't imagine they're long enough to allow your testers to work through every possible combination of functions in the software...or even a majority of them. Especially in a Virtual world, where the possibilities are almost endless. That being the case, the likliehood of at least one major bug appearing each week is very high. The end result is, of course, these "emergency patches." And the net result of that is:

1. More untested software released into the world, raising the probability of even more high-priority bugs.

2. Time taken away from development of your next planned release, meaning less functionality and/or more under-tested software in the subsequent version.

3. Residents experience downtime on a weekly basis while aforementioned patches are deployed.

Too-frequent releases has been the death of UO as well. This concept is more reflective of some "think outside of the box/move the needle" corporate bureaucrat than an actual Release Manager, Product Manager, or other software development professional -- any of whom would tell you that weekly releases are a bad idea. Let me ask...do they have a catchy name for this strategy yet? "Rolling development" might be a good one...

For God's sake, it's software. There are going to be bugs. Stand up to your management team and tell them so, and then come up with a real development and release plan so you can minimize the impact to your customers.

*stands up and applauds* bravo. thats what we've all been saying but you word it so eloquently

you hit the naIL ON THE HEAD why they need weekly updtaes do them bi monthly work the bugs before causing caouse on sl and allow sl customers to enjoy the service they pay for

*count the amount of topic about this*
*count the maximum people i ever saw in the preview grid*

Mmm.... there is something wrong here.



Please, be nice and try to write correctly.

Yes there is something wrong here!

Look in find (oh wait thats "search" now) sometime, theres like umpteen Lindens listed, this should be their job to beta test. Sure i love to see the beta grids and always go out of curiosity, but i dont have a clue what im looking for in bugs...its not up to the paying public they serve (yes i know thats a scarce verb nowdays) to do their (paying) jobs for them. They have the ability to beta test, and should do so for at least no less than 30 days before releasing, the beta grids are mostly used now as a marketing tool rather than a way to actually fix bugs.

Updates that take days out of a game that provides an income to any extent, should not be down this often and long for updates or bug fixes. What they need in LL is some leadership with some good old fashioned common sense...

Thank you for listening to my UMPTEENTH rant on this subject

Kerunix, you miss the point here.

a) The preview grid should be a beta, not an alpha.

b) Developers should not count on their paying public to provide their primary QA. When I'm spending my money and my time to play a game, and some portion of that time is already spent cooling my heels waiting for the grid to let me in, the last thing I want to do is donate more of my time to testing unbaked software so that the company can meet its unrealistic development cycle.

c) (note: this one is important) With weekly development cycles, no beta is going to be valid. There simply is not enough time to test a release of this magnitude in one week. By ANYONE.

Personally I think you all couldn't be more wrong. Release early, release often. The more frequent the updates the fewer new features are introduced and therefore the easier it is to track down bugs. Releasing more frequent smaller updates reduces complexity, improves stability.

Arguing that there isn't enough time to test new features in a single week is totally backwards. The way it works in reality is that smaller incremental changes are made that are easier to test. Besides, who's to say that a feature that was first developed in early May hasn't been undergoing testing since then only to be deployed in late July.

It really look like a beta, not an alpha.



They aren't, for some obvious reasons.)



You don't really know the development cycles in LL.
Do you really think they spend only 1 week to develop and test a new feature ?
I know some feature that are in developments since MONTHS ! (and, sadly, not released yet)

It's not like : "Ok, the maingrid is up and running, let's start a new development cycle for the next 7 days"


EDIT :



It was my argument a few days ago. (and it's still... i just don't want to repeat myself *too much*)

I think it's possible that with a smaller application and a limited development staff that might be true. But I think you understimate the ability of a "small," seemingly isolated code change to impact other areas of the software when you're talking about so much code. Fast releases negate the ability to do Regression testing for these types of issues.

People are replying on a technical standpoint. I know "dumbing it down" seems a bit patronizing for some of you guys... but c'mon. People come here for fun right? People don't necessarily want to get up to their elbows in mud dealing with this crap. They want a product that works. It's not working right now (and hasn't been for a good while) and the reasons it's not working 3/4 of us feel could have been easily avoided.

Patronizing other users for their lack of understanding, or for someone else's elaborate experience in the subject matter, doesn't really make a f**king difference. I don't care what you know or don't know. I don't care if you're education has you fit to run this game yourself. Simply? A service is broken, and its residents are antsy. You can't blame us for that. Similarly... your power is a paid service. If your power went out 4 times a week, you'd be pelted with excuses from the company...but you'd still be flustered.

I knew you were smart!

Every MMORPG I have played since MMORPG's hit the gaming world went offline once per week for maintenance and updates. It's part of the beast, it's normal. Actually, it's industry standard.

SL is down right now to fix what was referred to as "A MAJOR EXPLOIT". SL should be down right now ot fix it.

If people are so addicted to Second Life, that any moment where it is offline becomes a major life issue, then perhaps it's time to find a support group.

I had an idea here: /108/f5/116260/4.html

To requote it:

Ok. With all the things going on about the last update, and how bugs got into the release in-mass, Let's dicuss solutions to the idea of testing releases for problems before making them mainstream...

Aye. LL is kind of 'damned if they do, damned if they don't' right now. If they don't shutdown and the problem get's widespread abuse, they get accused of not doing anything. Now people are complaining that it's happening during a major event. No one is happy right now, and I sympatize.

The fact that major events happen often, and it's kind of hard to schedule around all of them for patches is beside the point. <shakes head> Aye. A better quality control could help.

How about this idea: What if we have three levels of SL: The Stable, Beta, and Alpha sites?

The idea is this. The stable is the 'real' SL world. Alpha is like the preview sites we have now.

Beta is a smaller, but usable world. It's run like the stable, but seperate. They give heavy discounts to land ownership tier, linden buying, and such. Nothing from Beta can enter Stable without special linden request and payment of a fee, and only if such a device is using tech that will work on Stable.

This would entice a large number of people to use the systems BEFORE they reach stable, allowing tons of time to see side effects of any changes. If a person does not login to beta for a reasonable amount of time during a month, they are removed and someone else allowed to come in. All applicants to beta must be approved. Etc...

Perhaps we could even give discounts/rewards to people who login to the Beta grid and provide bug reports for the main grid. <shrugs> It's open to debate wich would funnel the most manpower into beta testing.

Thoughts?

Just for the record I'll concede that it would be nice if they could find a way to apply updates without bringing the entire grid down for 12 hours (or whatever it is). It really should be possible to do the whole thing in about an hour or so. As for Rachel's last post, I think she's still stuck on the idea that each new feature is developed and tested in under 7 days.

Another point is that it is absolutely impossible to predict what changes may or may not effect something negatively in the full production environment. This isn't like adding a new smiley to the forums, each change/fix could possibly effect thousands of different things, and until an update is added to the "live" servers, it's 80% science and 20% "hope it works". That's the nature of a complex 3D interactive environment.

That, and not nearly enough people go to the preview grids, and many that do only go to "see the new feature" not actually truly "test" it.

You're correct, Sable...I don't know how long their actual development cycle is and I think I indicated that. But, if a release comes every 7 days, then your maximum time to develop on any given release is 7 days, unless you have multiple, separate development and QA groups completely dedicated to each planned release. Now add in the emergency patches to the existing production codebase which distract from development on future release(s)....

So, I can deduce that the QA cycle is probably at best 7 days after a code freeze. That this is inadequate is proven by the frequency with which we discover "major exploits" between releases.


*shrugs* But yes, I could be wrong.


As a side note, I also agree that major exploits require the grid to be brought down and the problem fixed in an emergency patch. I'm only suggesting that the accelerated releases are causing much of this, and perpetuating the problem. I have no issues with being down for releases and patches once a month or so, but every week is mostly preventable with a reasonable release cycle. And if I only have 5 hours a week to play and the grid is down for 3 of those, aren't I going to be even more ticked off than if I'm on for 20 hours a week?

Anyhoo, 'nuff said, at least by me.