stephane richard wrote:
well, since I don't have OS4 or anything to run it, it's sorta hard to do upfront coding per se. (except for the modules mentionned so far that could be coded with OS 3.1 (like aspell for example was mentionned).
AS far as my schedule goes, it can vary some as the rest of my days can vary just as much :-). Per week however I can probably easily put a good 20 to 30 hours of programming (maybe more too but just at different times on different days and different weeks). morning on some days, evening and late evening on others.
In the spirit of the KISS method, I would have to advise to keep things as independant as possible. As in if there's a "reasonable way" to do something without relying on 3rd party projects, I'd vote for that (provided it doesn't mean we double the length of the project :-). but for 5% to 10% faster time because we're using that 3rd party library or binding, to me that would be almost negligeable in my vote to not use the given library or binding. This way we're well independant and have 100% control over what the 3rd party lib or binding would have done. This ultimately always ends up in faster debugging time at the end, at least it's been my observations in my other programming projects.
Equation: If TimeSavedBy3rdParty > TimeToDebug3rdParty Then UseIt Else DoNot ;-)
Bear in mind there are some dependencies that are simply not a realistic option to re-write. We evidently need a [tc]sh, we're going to need a pthreads implementation, and other things. Wherever these come from that already exist, and hopefully work(!), will save significant amount of time over writing everything from scratch, unless I'm misunderstanding your point here?
OO as a large project has a good number of dependencies, none of which I think should arbitrarilt be 'replaced' because they're external. Virtually ever open source project has some 'external' dependencies. We should, however, if given the option, designate a specific version or version range for internal build work and porting efforts however, so as not to overly complicate (or break!) things..
Scott
"On Error Goto Hell" ;-)
Stephane Richard aka MystikShadows Software Developer (and AmigaOne wannabe ;-) srichard@adaworld.com
[snip]