SLDev-Traffic 15
sldev-traffic no 15
sldev discussions through June 8, 2007
Mute Visibility
Nicholaz Beresford created VWR-1017 to propose the ability to suppress objects owned by others. There was some discussion on list about what form muting might take. Tateru Nino pointed out that since physics are performed server-side, the viewer alone couldn't make objects phantom, which poses a problem when colliding with ghosts. Able Whitman suggested the ability to set a partial alpha on these objects and suppress touch, llTargetOmega rotations and texture display. There was some further discussion toward refining the concept, and Able Whitman has begun researching how object and texture display is initiated by the viewer, including posting the sequence of functions used in bringing an object into view on the viewer, but no firm plan or design for visibility muting yet exists.
Debian Etch i386 Builds
Paul Hampson announced that a user had performed i386 builds against his Debian patches. The result can be found at http://alzental-castle.de/~rd/SL/ - there is the added requirement of libcurl-openssl3, available from http://snapshot.debian.net
Debian Packaging
Paul Hampson stated his intention to get the SL viewer included in Debian's official release. Soft Noel (Linden) inquired about whether Debian would accept packages that quickly become outdated, and Paul pointed to the existence of the debian-volatile project, which now offers special case handling of exactly this type of fast moving target.
Bug Triage and Housekeeping
Rob Lanphier (Linden) made a request for help in sorting out an agenda for a bug triage Monday, June 11. Nicholaz Beresford and Able Whitman pitched in, and Dzonatas Sol moved the agenda to a static, undated label at Bug triage/Agenda so wiki users could watch the page.
As a reminder, on Mondays, bug triages happen at 3pm at Rob's office in Grasmere. If you're curious what goes into a triage, you can read the previous transcript at Bug_triage/2007-06-04.
CPU Detection
There was previous discussion about the best way to identify the CPU speed under Linux, with regard to issues VWR-940, VWR-962, VWR-975 and others. Dzonatas Sol posted code with the Linux version of the clock speed detection. The approach is similar to source's Windows version; that is to simply read the CPU cycle counter, sleep a few microseconds, read again, and scale to a one second interval.
Dzonatas also posts an updated patch against VWR-962, for CPU type detection, with Thomas Rowland helping out with a minor fix.
Contributing Code You Didn't Write
There was some speculation about the authorship of one piece of code under consideration for CPU speed detection, and what license it might fall under. Rob Lanphier (Linden) took the opportunity to post a strong reminder that any code not authored by the contributor should be noted as such. "Almost every software license out there (including MIT and BSD) have minimal requirements for acknowledgement." Treating code under a relatively unencumbered license as if were in the public domain is dangerous and potentially costly. Even code assumed to be in the public domain should be noted.
To be completely safe, Rob requests that third-party code be linked to in the JIRA, but not actually submitted as a patch.
Patch for VWR-176
Nicholaz Beresford posts a fix for VWR-176, a crash activated by a sequence of object manipulations sometimes performed by accident. The patch is attached to the JIRA issue.
Soft Linden
Soft Noel, your humble SLDev-Traffic editor, announced becoming a Linden this week.
To break the third-person schtick for a bit, thank you all so much for the congratulations on-list and off. Don't be shy about saying hi in-world, and if you're a signed contributor, consider friending Soft Linden so you're easier to harass with patch questions!
In response to the announcement, Nicholaz Beresford, aka "The Mad Patcher," got a bit silly with a terrible bit of Photoshoppery. The image is currently visible on his blog. Stick to coding, Nick. ;) And thanks again. :)