[opensource-dev] VWR-12984 Water flickers and disappears in patches (was: Stuff I'd like to see in the coming sprint)

[Boroondas Gupte]

 On 10/19/2010 04:41 PM, Oz Linden (Scott Lawrence) wrote:
>>
>>
>>         VWR-12984 <https://jira.secondlife.com/browse/VWR-12984>
>>         Water flickers and disappears in patches
>>
>>     * Very annoying and high-voted bug, especially apparent in skyboxes
>>     * Has been fixed in Snowglobe 1 half a year ago
>>     * Fix has been ported to Snowglobe 2 this summer (see SNOW-643
>>       <https://jira.secondlife.com/browse/SNOW-643>)
>>     * Merov ported the fix to viewer-development recently, so the
>>       code 'only' needs some review. (The change is huge, though.)
>>
>
> I'm not seeing this in Second Life 2.2.1 (212292) Second Life Development
Have you tried the repro? (I've put a platform back at
http://slurl.com/secondlife/Hippotropolis/220/164/1001, so the repro can
be performed again.) I can still reproduce this with current
viewer-development.

> are we sure it wasn't fixed by the other (and much simpler) water
> flicker fix?
I am sure. While the symptoms of VWR-12984
<https://jira.secondlife.com/browse/VWR-12984> and STORM-306
<https://jira.secondlife.com/browse/STORM-306> were similar, the causes
were very different and so are the fixes.

> In any event, nothing with a patch that large (and which affects code
> shared with the simulator) counts as "low hanging fruit", however
> important the problem is.
Fair enough, but it's certainly much lower hanging than before Merov
ported it.

> As I've mentioned before, there is an effort under way to split out
> the code that is shared between viewer and simulator into its own
> repository so that it can be better managed and evolved in a more
> coordinated way.  [...] My own opinion is that considering changes as
> large as this one that affects that code should be deferred until
> after that change.
Is this the case for this fix? While the amount of changed code is huge,
I'd assume it's mostly if not completely viewer-only parts that get
altered. Though, not knowing the server source, I cannot be certain.

I guess the indra/newview directory is viewer-only. I don't know about
indra/llrender, but I don't see any apparent reason why the server code
would rely on that.

> That should be visible in the open source within a month or so, I think.
Do we have any volunteers to forward-port the fix /again/ after that
time? And no, as much as I'd like to see this bug fixed, I'm not
volunteering for that task.

Cheers,
Boroondas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.secondlife.com/pipermail/opensource-dev/attachments/20101019/4efc2c93/attachment.htm