From max@duempel.org Tue Jan 25 11:06:40 2011 Date: Tue, 25 Jan 2011 11:06:40 +0100 From: Max Kellermann To: rkalman@hansaaut.hu Subject: LK8000 engine Message-ID: <20110125100640.GA31819@squirrel.blarg.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.20 (2009-06-14) Status: RO Content-Length: 4707 Lines: 95 Hi Kalman, I saw you just joined the LK8000 project as a developer. I appreciate that, because it means you contribute to the whole ecosystem. Your topology label fix was done by us on XCSoar half a year ago: http://git.xcsoar.org/cgit/max/xcsoar.git/commit/?id=151a90f5 This is just yet another example of wasted time. There are countless other examples. We should better join forces, and not fix the same bugs on XCSoar and then on LK8000 again. Below, you will find an email I sent to Mateusz last week. Might be interesting for you, too. Also note my email on the XCSoar developer mailing list: http://sourceforge.net/mailarchive/forum.php?thread_name=20110120093612.GA14648%40squirrel.blarg.de&forum_name=xcsoar-devel Max ----- Forwarded message from Max Kellermann ----- From: Max Kellermann Date: Tue, 18 Jan 2011 12:14:22 +0100 To: mateusz.pusz@gmail.com Subject: LK8000 engine User-Agent: Mutt/1.5.20 (2009-06-14) Hi Mateusz, Tobias told me you talked to him about the LK8000 engine. You might find this chat log with Paolo interesting (Dec 24th 2010): 02:55 <@cirrus> and until the release, your code will be secret? 02:55 <@cirrus> I'm curious why you keep it secret, because we attract a lot of developers with our public source repository. 02:56 <@cirrus> well you can always use our task engine. it's excellent, really. jmw's masterpiece. 02:56 < paolo> the source code of LK makes people run away, just like 5.,2.4 , thats why 02:56 <@_Turbo_> paolo: then why don't you port your interface on top of 6.0?! 02:57 <@_Turbo_> after all 6.0 is much easier to modify and porting it will obviously take time but then you would have all the advantages 02:57 < paolo> its not only an interface... you know that. 02:58 <@cirrus> it is less time to port LK8000 changes on top of 6.0, than duplicating all the work we did (many man-years). 02:58 <@_Turbo_> but still... having all your features shared might be beneficial for both projects 02:59 < paolo> john declared that he refused to use LK interface at all in july 2009 02:59 <@cirrus> I would estimate we added around 7 man-years of work since 5.2.4 (jmw, _Turbo_, rob, me and several others working nearly full-time on it). 02:59 < paolo> this is the reason why we forked, also 03:00 <@_Turbo_> yes, but that doesn't prevent you from using the 6.0 code base 03:00 <@cirrus> it's open source! forking doesn't mean diverging trees. 03:01 < paolo> I think I shall wait for some months to see that nothing is hanging inside xcs 6 before thinking abut merging code 03:01 <@cirrus> the 6.0 branch is frozen. nothing will happen there except for bug fixes. [...] 03:24 <@cirrus> the new engine is paying off then. 03:24 <@cirrus> the whole potential of the new engine isn't used currently. 03:25 < paolo> engine? 03:25 < paolo> what is that about 03:25 <@cirrus> task calculation and the whole xcsoar backend. 03:26 <@cirrus> it is so well-designed, it took jmw only very little time to implement olc2011 rules - by combining two existing classes. 03:26 <@cirrus> OO at its best! 03:27 <@cirrus> I'll be curious how you implement that with the old xcsoar 5 code base. quite hard to do. 03:27 < paolo> implement what? 03:27 <@cirrus> olc2011 03:28 < paolo> I dont now 03:28 <@cirrus> aka olc plus 03:28 < paolo> I dont even know what it is 03:28 <@cirrus> the olc code in lk8000 is obsolete 03:28 < paolo> thats for sure 03:28 <@cirrus> the results are useless because olc has new rules now 03:29 < paolo> I shall check then 03:30 < paolo> I dont have any hurry on these matters.. most people is asking about basic functionalities for safety.. 03:30 < paolo> and I cannot do everything of course... 03:31 <@cirrus> we know that, that's why we suggest you build on top of xcsoar6 instead of 5. 03:31 <@cirrus> you cannot ever catch up with xcsoar as long as you're alone 03:32 <@cirrus> without merging, that is. Seriously, LK8000 has already fallen behind XCSoar 6 (regarding the engine). Life is too short, you two shouldn't waste your precious time on problems we have already solved. Do you you really want to re-do a man-decade of work? I would like to share more code between XCSoar and LK8000. If you rebase LK8000 on top of XCSoar 6, we could easily share things like device drivers, calculation algorithms, rendering optimizations, the Android code and so on. Just by invoking "git cherry-pick". (By the way, Paolo keeps telling he forked because John wouldn't accept his UI patches - see for yourself: http://xcsoar.1045713.n5.nabble.com/A-few-general-questions-on-development-td2650276.html#a2650278) Max ----- End forwarded message -----