On Jan 8, 2008 12:49 PM, Mike Caron <[EMAIL PROTECTED]> wrote: > On Jan 7, 2008 6:43 PM, James Paige <[EMAIL PROTECTED]> wrote: > > On Tue, Jan 08, 2008 at 11:51:19AM +1300, Ralph Versteegen wrote: > > > On Jan 7, 2008 7:35 PM, James Paige <[EMAIL PROTECTED]> wrote: > > > > On Sun, Jan 06, 2008 at 10:06:32PM -0800, [EMAIL PROTECTED] wrote: > > > > > pkmnfrk > > > > > 2008-01-06 22:06:32 -0800 (Sun, 06 Jan 2008) > > > > > 837 > > > > > FB 0.18.3 compatability > > > > > > > > > > Ouch, I've touched just about every source file in the tree -_-. > > > > > Sorry if I broke your commit, but the changes should be easy to merge > > > > > in. > > > > > > > > > > Basically, any global string that used to be "whatever$", is now just > > > > > "whatever". Any global integer "foo%" is now "foo". Apparently, > > > > > Extern doesn't like type suffixes in 0.18.3, and it's backwards > > > > > compatible not to use them, so they don't any more. > > > > > > > > > > I am fairly certain that I got them all, but it is possible I missed > > > > > one. Theoretically, if I did, then this code would compile just fine, > > > > > but wouldn't work: > > > > > > > > > > sub foo() > > > > > game$ = "newgame" > > > > > end sub > > > > > > > > > > since game$ is now game, the compile would think that game$ is a new > > > > > variable, and be perfectly happy. However, game would not be set to > > > > > "newgame", potentially causing problems. I doubt that we have any > > > > > such cases, though. > > > > > --- > > > > > > > > Bravo! This actually takes care of the second-worst part of making the > > > > OHR -lang fb compatable. (the worst, IMHO, being gosubbery) > > > > > > Remind me if I've forgotten, but what's wrong with our gosubs? > > > > I was under the impression that our nifty GOSUB/RETRACE hack didn't work > > with -lang fb ... but I could be mistaken. > > > > We have no idea, since we can't compile with it. > > But, there was some question about it's reliability around loop > constructs. I did an experiment, which is archived somewhere on the > mailing list, but I forget what its results were.
I just tested it, our gosubs seem to work under -lang fb, subject to the rules for what you can't do in a gosub block. gosub.el also seems to still work for checking for gosub safety (under Windows at least). > > Either way, longjmp() is a well documented function in the Standard C > Library, so other than the compiler generating truly bizzare code > which somehow renders longjmp() ineffective, then it should still > work. You forget, I wrote a new GOSUB hack which uses assembly. > > > > -- > Mike Caron > Tale of the Cave > http://taleofthecave.com _______________________________________________ Ohrrpgce mailing list ohrrpgce@lists.motherhamster.org http://lists.motherhamster.org/listinfo.cgi/ohrrpgce-motherhamster.org