[plt-scheme] Announce: SQLI/SQLD PLaneT package
I'll be in trouble I think going the FFI route because of one important
implementation detail in sqlid drivers:
They use OS threads to start queries. This is a necessity, because
otherwise, querying the database will stop the entire PLT Scheme
process, including all threads, which is absolutely not desired
behavior.
--Hans
pedro pinto schreef:
> You may also want to explore the FFI route. I am not sure what kind of
> changes that would force on your code but it should make distribution
> easier.
> -pp
>
>
>
> On 1/24/06, *Hans Oesterholt-Dijkema* <hdnews at gawab.com
> <mailto:hdnews at gawab.com>> wrote:
>
> I see what the problem is. My binaries are linked against
> libmzsch300_000.dll, and not
> against libmzsch301_00.dll. I guess this is why with Gtk+,
> although the
> version
> number increases, the libraries are still named e.g.
> "lib-gtk-win32-2.0-0.dll" instead
> of "lib-gtk-win32-2.8-9".
>
> Is it not a (good) idea to use a different distribution scheme, e.g.,
> like UNIX, where
> libxyz.so is actually a link to libxyz.so.1, which is a link to
> libxyz.so.1.3.9. If an
> application links against libxyz.so.1, it will also work with
> libxyz.so.1.4.1 and only
> break when the library goes up to libxyz.so.2.
>
> So on WIN32, libmzsch3.dll should be there to link against. And
> for more
> granularity:
> libmzsch30.dll, libmzsch300.dll, libmzsch300_000.dll.
>
> Is it an idea to support this?
>
> --Hans
>
>
>
>
> Hans Oesterholt-Dijkema schreef:
> > By the way, what is the error (message) you get?
> >
> > pedro pinto schreef:
> >> You may already know this but just in case, I noticed your
> >> c-sqld-*.dll files have a dependency on v300 of the PLT dlls so
> they
> >> will not work with 301 even if the drivers are present. I guess
> that
> >> is one of the pitfalls of distributing precompiled files.
> >>
> >> Take care,
> >> -pp
> >>
> >>
> >>
> >> On 1/8/06, *Hans Oesterholt-Dijkema* <hdnews at gawab.com
> <mailto:hdnews at gawab.com>
> >> <mailto:hdnews at gawab.com <mailto:hdnews at gawab.com>>> wrote:
> >>
> >> pedro pinto schreef:
> >>
> >> > Mathew's delay load suggestion seems like the best bet then.
> >> >
> >> It is, I've actually implemented such a scheme. However,
> without
> >> the DELAYLOAD option, because I don't have DelayImp.Lib
> >> installed on my system (I'm using microsofs freeware compiler).
> >>
> >> best whishes,
> >> Hans
> >>
> >>
> >>
> >
> > _________________________________________________
> > For list-related administrative tasks:
> > http://list.cs.brown.edu/mailman/listinfo/plt-scheme
> <http://list.cs.brown.edu/mailman/listinfo/plt-scheme>
>
>