Subject | Re: [IB-Architect] Did anyone get a win32 compile going? |
---|---|
Author | ib-architect@mers.com |
Post date | 2000-07-30T17:48:10Z |
Mark O'Donohue wrote:
Well, partially. I sure couldn't use the supplied makefiles,
batchfiles, sed-scripts as-was, but I have managed to build (most of)
it. The server starts but doesn't work, could be that I have compiled
it with the macro SUPERSERVER defined, since it spawns off an infinite
number of threads and basically doesn't work. But it compiles. :-/
On GNU/Linux you probably could (partially) use the build scripts? In
Win32; forget it! You'd need proprietary tools (MKS toolkit - since
ksh is used, Borland resource compiler, sed - which many Win32
developers don't have a copy of or even a clue of what it is). I've
seen something about (limited?) success using the Cygwin tools and
bash in one of the newgroups, but I can't verify it. I don't know how
the build went on GNU/Linux, but using MSVC6/Win32 it produces a
*gazillion* (and I actually don't think that's an overstatement)
compiler warnings. I've edited files for hours and hours to nail a lot
of those little buggers (including some bugfixes along with'em), but
without a place to post patches I'm afraid they will go to dev/null...
:-(
One of the most frustrating things about it, except that I needed to
D/L the binary version to get the src to compile, was that the
directory structure isn't very logically organized, to put it mildly.
My personal preference would be one directory for each
library/subsystem, and the same for the executables. As it is now, it
seems to "borrow" heavily from each others directories, having
circular dependencies, and namig the directories and files to dorky
names that have no place in a world where (MS-DOS) 8.3 names have *no
place* whatsoever (e.g. to name the threading stuff "thd" is a
stretch. "thd" is probably pronaunced "thud", and that was the
not-so-nice nickname of a military aeroplane once :-) ).
Now that I spilled my guts over this, it has taken me since the
release of the code until a few hours ago until I managed to compile
most of it. Most of it, since Win32 resources can't be compiled
without Borlands resource compiler. It might be free to D/L (if you
register and possibly give up the rights to your name, mail-address to
spammers, and your life), but from the way Borland/Bigprices/Mr.
Drooler (or Fooler or whatever he's supposed to be) has handled this
whole issue I'm quite confident in saying that I don't want anything
to do with them or their products, ever. I will try to post working
.rc files to mers.interbase.build (if binary (zipped) files are
accepted), but it could take a while.
Besides that, I've got working command line tools (well, I tested gpre
and it worked), but the server isn't working just yet.
/Mike
_______________________________________________
Ib-architect mailing list
Ib-architect@...
http://mers.com/mailman/listinfo/ib-architect
>I was just wondering if anyone got the win32 distribution compiled fromUsing Win32, MSVC6 and a sed compiled for Win32:
>source.
Well, partially. I sure couldn't use the supplied makefiles,
batchfiles, sed-scripts as-was, but I have managed to build (most of)
it. The server starts but doesn't work, could be that I have compiled
it with the macro SUPERSERVER defined, since it spawns off an infinite
number of threads and basically doesn't work. But it compiles. :-/
>If so I would be interested to see how the problems were similar/different toI'd say that it's a world of difference! :-)
>the linux ones.
On GNU/Linux you probably could (partially) use the build scripts? In
Win32; forget it! You'd need proprietary tools (MKS toolkit - since
ksh is used, Borland resource compiler, sed - which many Win32
developers don't have a copy of or even a clue of what it is). I've
seen something about (limited?) success using the Cygwin tools and
bash in one of the newgroups, but I can't verify it. I don't know how
the build went on GNU/Linux, but using MSVC6/Win32 it produces a
*gazillion* (and I actually don't think that's an overstatement)
compiler warnings. I've edited files for hours and hours to nail a lot
of those little buggers (including some bugfixes along with'em), but
without a place to post patches I'm afraid they will go to dev/null...
:-(
One of the most frustrating things about it, except that I needed to
D/L the binary version to get the src to compile, was that the
directory structure isn't very logically organized, to put it mildly.
My personal preference would be one directory for each
library/subsystem, and the same for the executables. As it is now, it
seems to "borrow" heavily from each others directories, having
circular dependencies, and namig the directories and files to dorky
names that have no place in a world where (MS-DOS) 8.3 names have *no
place* whatsoever (e.g. to name the threading stuff "thd" is a
stretch. "thd" is probably pronaunced "thud", and that was the
not-so-nice nickname of a military aeroplane once :-) ).
Now that I spilled my guts over this, it has taken me since the
release of the code until a few hours ago until I managed to compile
most of it. Most of it, since Win32 resources can't be compiled
without Borlands resource compiler. It might be free to D/L (if you
register and possibly give up the rights to your name, mail-address to
spammers, and your life), but from the way Borland/Bigprices/Mr.
Drooler (or Fooler or whatever he's supposed to be) has handled this
whole issue I'm quite confident in saying that I don't want anything
to do with them or their products, ever. I will try to post working
.rc files to mers.interbase.build (if binary (zipped) files are
accepted), but it could take a while.
Besides that, I've got working command line tools (well, I tested gpre
and it worked), but the server isn't working just yet.
/Mike
_______________________________________________
Ib-architect mailing list
Ib-architect@...
http://mers.com/mailman/listinfo/ib-architect