Hey Dennis, On 22 Sep 2008, at 16:24, Dennis Schridde wrote:
> Am Montag, 22. September 2008 15:29:09 schrieb Freddie Witherden: >> Hi Dennis, > Hello Fred! > >> On 22 Sep 2008, at 13:06, Dennis Schridde wrote: >>>> Try playing 2.1_beta4, it is quite poor so far as releases go. >>>> 2.1_beta5 is not going to go down well with people either if we >>>> break >>>> their save games (will probably cause them not to upgrade). >>> >>> That reminds me that we need some way to have backward compatibility >>> somehow. >>> Savegame-wise and whatever else might need it. At least so much as >>> there is a >>> simple way of conversion. >> >> The main problem is game.c. It makes creating new save game versions >> very difficult. Hence, the tagfile and SQLite proposals for save >> games. > It's mainly copying the code, changing the version numbers, and > changing the > parts which are different, right? More complex than that, I estimate that between 400-500 lines are required per 'version'. > How far are the tagfile and database ideas? Any progress there? I > know the > tagfiles basically seem got stuck after the early phase of > implementing the > framework functions... Not too far along, they would both require weeks of work. >>> Speaking of conversion: The only thing that makes =2.1_beta4 games >>> not load in >>> >>>> 2.1_beta4 is that the static gateway and zone information is >>>> missing? Can't >>> >>> we just copy that from the original map again? (In a conversion >>> step, maybe as >>> an external tool if necessary.) >> >> It would be a lot of effort that would only be useful/used in beta5. >> Furthermore it would need a lot of bug-checking, perhaps more so than >> getting trunk 100% stable. > Better than leaving a out a release and letting the ship sink in the > dream > that the next release would come anywhere "soon". To support save games with and without zones would be a massive undertaking. It would be paramount to adding a large amount of relatively untested code to a beta release, written under a tight time constraint, that would only ever be used in 2.1. Writing code for a single, already outdated release is foolhardy. This is not a good use of developer time -- which could be better spent on 2.2 -- ensuring that we never get into this situation again. >>> You may say, blahblah, it's just 2.1_beta5 and does not offer as >>> many great >>> things as 2.2 (which no one can tell me is going to arrive in the >>> next week or >>> even month). But we need some backward compatibility layer anyway, >>> so we could >>> use this as a testbed... >> >> That is what game.c provides, albeit badly. > So even if game.c sucks, why not use what we have for now? > And do not forget to implement a better compatibility layer for 2.2+? The amount of work makes in prohibitive. Regards, Freddie. _______________________________________________ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
