On Fri, 24 Jul 2015 21:56:28 +0200 win...@genial.ms wrote: > it crossed my mind that some time ago people were complaining > about the Mac builds not working.
AFAICT that was due to Apple's somewhat adversarial treatment of Java. There were follow ups saying "install the <some-version-of-Java> package and it works". I have definitely sighted 0.11.3 working on Macs. > Another thought I had was, that the codebase collected many > compatibility hacks from the overly long 0.10.x series. > I did a search and there are about 243 hacks for 0.10.x and > about 141 hacks for 0.11.x already. > The 0.11.x series already had 3 patch releases for 0.11.0, which > is in line with how it was done until 0.9.x. I think that is just a historical accident. > We have new features implemented in addition to all the bugfixes, > which also suggests increasing the minor version. > So, I would like going to 0.12.0 directly. > After the release we could drop 0.10.x compatibility and get rid > of 2/3 of the hacks. I strongly disagree. Since I joined the project the FreeCol project, the minor version has only changed when we make an incompatible change to the save game format, in the sense of no longer being able to import games written by the previous minor version. ATM we still support the 0.10.x format, so there is no urgency to bump the minor version. Furthermore we have recent valid bug reports with 0.10.x-originated games attached, so there may be user impact to consider. As to the number of accumulated backwards compatibility hacks, the important question is whether they are impeding progress or not. Speaking as the main instigator of the last minor version bump, believe me, if you think the 0.10.x compatibility code is annoying, be thankful you did not have to work with the 0.9.x stuff! We are nowhere nearly as badly off as things were in 0.10.6+. However in the end what prompted the last bump was the existence of two clusters of bugs that were outright unfixable without an incompatible change to the save game format, not mere developer discomfort. This lead to the tile caching and equipment->role changes. Note that there is reporting bias in the numbers. I only started to add the @compat <release> annotations late in the 0.10.x series, but since then I have been fairly meticulous about them. So the number of 0.10.x hacks is probably lower than actually exist, whereas the 0.11.x ones are close to accurate but include trivial changes. Returning then to the question of progress --- I just did a quick sample of the annotations, and found they were highly concentrated in the serialization code, which is not a huge surprise. This is not an area needing a lot of work ATM IMHO, given that it was almost totally revised for 0.11.0. So I have to ask if there is a specific problem that the compatibility code is causing? >[Caleb] > FreeCol should take advantage of any features of Java 1.7 that it can. Quite a few Java 7 improvements have been made (for example use of the diamond operator, variables in try blocks). Do you have anything in mind that we are not doing? Cheers, Mike Pope
pgpJWpcw8S3mF.pgp
Description: OpenPGP digital signature
------------------------------------------------------------------------------
_______________________________________________ Freecol-developers mailing list Freecol-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freecol-developers
