Preparing FB 0.24

[rchockxm]

nice
wait for a long time.

thx

[albert]

I've been having the same problem as VANYA. ( Been having it since FB 0.20 )

The examples don't all compile out of the box. And the GMP example says Libgmp-3.dll is missing from you computer.

[MichaelW]

For the GMP example, gmp.bi specifies the import library libgmp.dll.a, and that import library specifies libgmp-3.dll. The FreeBASIC distribution cannot include all of the various libraries, so the solution is to build or otherwise obtain the binaries for a current library, and create or otherwise obtain a matching header file, and if the binary is a DLL a matching import library.

[creek23]

Is C-emitter shipping with 0.24? Is it anywhere near stable?

~creek23

[fxm]

I also hope that there will be the same features (inheritance, C-emitter) as the the Daily-Build at freebasic-portal.de, otherwise I'll stay with it!

[AGS]

I've posted the bindings for pcre 8.31 in the external libraries section (.bi and .def).
Unfortunately I cannot get the allegro bindings done in time for the release.

[TESLACOIL]

is there a specificish road map for FreeBASIC ? or are the developers just battling along as it where

[anonymous1337]

TESLACOIL:

http://fbc.svn.sourceforge.net/viewvc/f ... iew=markup

[Imortis]

This is the latest one:
http://fbc.git.sourceforge.net/git/gitw ... xt;hb=HEAD

[TESLACOIL]

thx, i always wanted to write a speedy simple language, just never got round to it...for me creating sentient AI is trivial by comparison,lol. Interesting to see whats going on behind the scenes.

...i do love a spaghetti monster in the morning

[counting_pine]

Hi dkl, I note the changes at http://fbc.git.sf.net/git/gitweb.cgi?p= ... ;h=135ff74.
It was a little out of the blue, though the basic "make && sudo make install" commands still seem to work fine.
Would you mind explaining more about the rationale(s) for it?

[dkl]

Yea, I figured it was unexpected (and perhaps even annoying) to have the layout change again, but luckily renaming isn't a problem with Git, and I hope we can stay with the way things are now.

Well, the src/ layout with target-specific subdirectories (as used in 0.23) is actually pretty good, because it allows module selection via wildcards. That makes the makefile much simpler and also allows compiling manually without a makefile. I originally got rid of it because there was a lot of duplicate code in the target-specific subdirectories, but now that's taken care of by splitting common code out of the offending modules.

Furthermore, the makefile doesn't use a new/ build directory anymore, but builds into the current directory, meaning either the source tree or a separate build directory from where "make -f .../path/to/makefile" was launched. That's no big difference, it just makes the source tree layout and build behaviour match the way FB works: You have one fbc and at least one but potentially multiple
target-specific rtlib/gfxlib2 builds.

Additionally, the headers aren't copied during the build process anymore; this is something I disliked a lot, because it's unnecessary, and also annoyingly slow on Windows. Now the makefile still recreates most of the needed directories to satisfy the normal or standalone layout, except for the headers, which sit in place for the standalone layout. Testing the headers with a normal fbc build without installing it now requires doing -i inc/, but I think there actually rarely is a need for it. This way we have standalone builds working directly with the source tree, and even though normal builds won't, those can be installed
somewhere via "make install", or -i inc/ can be used. All in all I think it's a good compromise.

[dkl]

I'm all setup for making the release, though currently I'm trying to solve a I/O lockup problem with the Unix rtlib initialization code that for some reason showed up with musl libc. I have seen something similar before with normal FB+glibc programs on very rare occasions, but with musl it seems to be a more persistent problem...

[dkl]

Allright, I'm pretty much done building & packaging everything. I'm going to do some more testing and then start uploading later tonight or tomorrow. (PS: if someone were trying to recreate the release, here's some notes on it)

What do you think are the most note worthy changes to point out in the announcement? The new inheritance keywords are important for sure, ThreadCall too, and also massive header updates (especially GTK+...), and many bug fixes (e.g. in the C emitter)...

[relsoft]

Are virtual classes implemented?