On Sat, Jul 5, 2008 at 10:43 AM, Robert Millan <[EMAIL PROTECTED]> wrote: > > On Sat, Jul 05, 2008 at 10:10:28AM -0400, Jim Park wrote: > > Clytie pretty much has it right about what's going on with NSIS and Unicode > > support. When we were looking for Unicode support in an Open Source > > installer package, we looked into NSIS and found no desire to support > > Unicode on the part of the developers. After two months of work to get > > Unicode support into the NSIS codebase and offering it to the NSIS team, I > > thought they'd take it but to my surprise they didn't. They didn't like > > certain things about my implementation and held it at that. They talked > > about how they'd like to do it better. I said, "Fine. Here's my code, you > > can use it to get started." It's been ten months and if you look at the > > latest posts, they are still talking about how they can do better. I don't > > think they've actually done any development towards Unicode support. (I > > hope I am wrong.) > Usually, when you want to get something merged in a project, you're expected > to be the one who puts the effort into adapting it to give it the shape > they want. Anyway, I don't know all the ins and outs; this was just a > general advice.
When you are in charge of an open source project and someone adds a new feature and offers up the patch as well, I think it's up to you to incorporate it into the project that is if you think it's worth it. If you don't want it, no matter how much the person bends backwards, you aren't going to add it. But I appreciate the advice. > For win32-loader, depending on two versions of nsis and providing two separate > builds adds a maintenance burden that I'm not willing to bear. What we would > really need is for utf-16 to be supported _but_ without making it mandatory > to use it in all strings. So if just some languages are utf-16 and Windows 98 > doesn't support that, we can enable them conditionaly in runtime. Does your > current code allow this? No, the current code does not allow this. I imagine that would be a lot of work. Quite frankly, my development group's needs are met right now. I'm not going to get any support from my team if I go spend more months on this project. I did two months of fulltime on this project. I'm not going to be able to convince anyone I need more time. I've got a ton of other work to do. I also don't think Win9x is relevant enough to spend my free time on it. And it will be more irrelevant as time goes on. Logistically, even if I wanted to do it, with the time constraints I have, you will have to wait at least a year. The amount of code change to get the Unicode support into NSIS by sheer number of lines is many more times than the amount of change the main NSIS went through over the whole of last year. You are talking about reading through 80,000 lines of C code and making changes to make the Unicode version work, not to mention the conversion of NSH files and language files to Unicode versions which sometimes was not a straight change due to systems calls or file access. But still it was simpler than any of the other options and I'm willing to bet that if I had to start all over again with a transparent Unicode / Win9x support as my goal, I would not have been able to finish it in the time I had. But if someone leverages off what I've already done, maybe they can do it a bit faster? I can imagine something quick and dirty like creating two exehead's for the installer that switches depending on the OS. Even this change will take some time to do and no guarantee that the NSIS group will take it. The scarier part of this change is that it will fork the code so much that I don't think I'll be able to maintain the code to be in sync with the main NSIS. It would probably end up being a forked and frozen version. (Frozen usually means abandoned in the open source world.) So it really has to be done by the main NSIS guys with the committment that they will support it going forward. Honestly, may be you should try to convince the main NSIS developers to get Unicode support a higher priority. I think they do want to do transparent Win9x and Unicode support. If they did that, I'd probably get our guys to use that project as well. - Jim -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
