2017 IMA: WG 31 TMFFail succeeds. The benefits of problem analysis

TMFFail succeeds. The benefits of problem analysisFinal report on the TMFFail project.  The project developed a  test object/protocol that could reliably replicate the problem.  The project was publicized in a blog article.  It attracted tests and expert opinion.  A fix was developed by one of the OS Devs and placed in the OS 0.9 master.TMFFailThe defect appears in two prims on the rightClick image to enlargeBackgroundEarly in June, I wrote about a problem that I called TMFFailThe problem: when a single texture is applied to multiple faces of an object,sometimes a few faces don’t change.IMA Test WG: Problem analysis tests, objects, protocols: Example: TMFFailI suggested that such tests could be a useful contribution by IMA WG31IMA WG31 Test Management work group is organizedResultsI got a number of helpful responses. The outcome was that Ubit, an OS Dev, found a way to fix the problem.After the break: details of the findings and comments on the benefits of test objects and protocols.*******************************************************************(More after the break, scroll down!)********************************************************************** Metaverse events, current and upcomingMatariki: Pleiades Short Poetry Workshop, Tue 6/27, 7pm SLTNgā mihi o te tau hou: This week the Short Poetry Workshop celebrates the Māori New Year – Matariki. The Matariki star cluster is known as Pleiades. Guest host Ngagpa Writer will introduce the Pleiades which is a 7-line poem. Each line of the poem starts with the same first letter of the title word. So join us at the Short Poetry Library to celebrate Matariki and learn, write and share the Pleiades.http://slurl.com/secondlife/Cookie/160/134/21/ **************Journal of Virtual Studies, Vol. 8, No. 2IMA Blog3D web-worlds of the browser (G+ community)Public service web-world by CybaLounge. Virtual worlders, this one is for usOSCC presentation videos availableCRASH! Site, the musical under development by people like usMal Burns: Inworld ReviewTry a browser-based virtual worldSelby’s daily news collection:  Virtual worlds are real   Virtual Outworlding news collection Events on the Hypergrid: HYPEventsEntertainment on the HypergridHYPERGRID EVENTSDestination guide: OpenSimWorldHyperica, Directory of Opensimulator hypergrid destinationsOpensim AAM Virtual PerformersWELCOME TO SEANCHAI LIBRARIESVirtual world G+ communities:  Communities virtual worlds G+MOSES software developmentIn a virtual world, your deodorant never fails.***********************************************CommentsThis investigation was only a pilot study and thus not the rigorous investigation that would be needed for publishable results.  The objective, however, was to find a fix for a persistent  problem.  As it turned out, a pilot study was sufficient to achieve that objective. (Translation: If I had needed a publication I would have had to put a lot more work into it.)Collected comments and suggestionsAbout a dozen comments indicated that the problem was well known.One person had seen it in Second Life, but not regularly.The opinion was unanimous that the basic problem was lost packets.There were differences of opinion as to what was affected.  The loss might affect the internal (storable) description of the object.or just the textures applied by the viewer. Or both. (My observations suggest both, but most often in the internal description.)Packet loss (network congestion) explains the erratic nature of the problem.The value of the test object and protocolErratic problems can be hard to work with because they are not easily reproduced.A major benefit of the test object here was to make the problem reproducible.No one can fix a problem without being able to reproduce it.  And if you can’t reproduce a problem, you can’t tell when you fixed it.A second benefit of this test object was that the test was easily duplicated by others.As a result, we got a lot of reports very quickly.We were able to determine easily that Digiworldz did not have the problem.Digiworldz did not know what fixed the problem.But the test object on Digiworldz did show that the problem was fixable.  The value of publicity for public problemsPublic reporting on the problem got a number of people to test it.   I did not test it in many places, but others did and gave me reports.  In a formal study, I would have recorded the reports and authors.  Here I just summarize the reports and acknowledge contributors.Results and conclusionsThe problem is corrected in OS 0.9 master  (Source Code Repository)The problem was found in OS 0.8 and 0.9Tests on Halcyon and Second Life did not show the fail.I have a report from one person who had seen the effect in Second Life.But my tests with the test object did not show the fail.  The effect is only a problem if it occurs often enough to bother builders.Lessons learnedIf we understand the problem to be packets lost, we can’t fix it.We can’t stop packet-loss on the internet.All software that handles packets from the internet deals with packet-loss.And everyone sees the results of packet-loss once in a while.There is a problem when consequences affect the user too often.That happens when the software does not handle the packet loss well enough.Are we finished?  No. We are finished when the test object and protocol are archived at IMA.And the report should be in some safe place, too.The object and test procedure may be needed for future QA tests.Since this was not a formal study, this report will do.Possibly this blog is safe enough.But I think we will want to create an IMA internal publications resource.  And any report should carry a link to object and protocol in the IMA archive.I will suggest that IMA set up an OAR archive for test objects.The objects would carry the relevant protocols on notecards in their contents.Suggested action items for IMA staffArrange for a WG that manages archiving and curating products of IMA WGsLibrarians and academics might be especially interested in this activityThanks for help:UbitShawn MaloneyDon SmithTerry FordBill BlightFred BekhusenFreaky TechBobCsirhc Tribe**********************************************************************************LicenseAll original text in this blog is CC By:Use as you please with attribution: web link to the original.All images without attribution in this blog are CC0: public domain.*********************************News and NotesThe Hypergrid WIP ShowThe Hypergrid WIP is a one hour “show & tell” of works in progress. Everyone is invited.  Building, scripting, entertainment– whatever you are working on.Selby may capture video of presentations in voice, for posting on YouTube.The WIP show normally meets on 2 Sundays a month at 12 pm California time.The show meets at the Pandora location on the second Sunday But will

Click here to read the complete story