[Texmacs-dev] Re: C++'s frustrations

[Henri Lesourd]

Abdelrazak Younes wrote:

> Henri Lesourd wrote:
>
> > Thus there is a problem: we need to go into the
> > business of telling plugins implementors which
> > tool to use.
>
>
> No need to tell, just state it on your "plugin development" web page, 
> that's all. Choosing a compiler is the job of the plugin developer, 
> not yours.
Right: except if I put him in a situation where he
cannot really decide, because I myself don't know
which compiler will be used to compile TeXmacs on
each of the platforms ;-(...

But another post discussing the issue suggested
one should be able to incorporate the compiler
info in the binary, then at least we could test
for incompatibilities.

It seems we are coming close to the definition
of an acceptable (if not perfect) solution, after
all these discussions :-)...


> So just standardize on one compiler (MSVC on Windows and gcc > 3.3) on 
> Unix) and be done with it.

Annoying.


> I don't think (read I am pretty sure) that the user will encounter any 
> problem.

As soon as the user can download a plugin independently,
and as soon as there is no explicit synchronization
between the plugin developer and TeXmacs, it seems
the user can download incompatibly compiled plugins,
or compile such plugins from source with the wrong
compiler...


> And as Josef said, inside TeXmacs, you can check the compiler used 
> before loading the plugins and if the plugin doesn't match the 
> requirement just display an informative error box saying why the 
> plugin cannot be loaded.
That seems to be the appropriately clean
approach, right.