|
Boost : |
From: rogeeff (rogeeff_at_[hidden])
Date: 2001-12-24 10:26:02
--- In boost_at_y..., Kostya Altukhov <kostya_at_s...> wrote:
> rogeeff wrote:
>
> >>All macro solutions require recompilation, therefore they
> >>are not acceptable for real-world needs.
> >>Everything that requires recompilation is not acceptable.
> >>
> >Macros are used only as a helper facility for format *definition*.
> >You can always have separate mechanism to support runtime cjange
of
> >the format.
> >
> It should be possible to update format definitions and add new
> format definitions without recompilation.
>
> It is often not acceptable if translating messages to another
language
> requires recompilation of the executable. However, translating
> messages often requires modifications in format definitions,
> such as changing parameter order.
>
> That's why in many international applications format definitions
> are stored in separate places (eg. text files), and it is possible
> to modify these text files without updating executable.
>
> I see no way of supporting this approach using streams
> without something like submitted formatter.
You can always implement runtime format "storage" as a dynamic
library. And you won't need an executable recompilation.
>
> Best wishes,
> Kostya
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk