Russell Yanofsky wrote:
>Branko Čibej wrote:
>
>
>>Generating project files that you know won't work isn't
>>right.
>>
>>
>
>The resulting project file will work if the user later installs Perl 5.6.
>And anyway, I think generating a project file that might not work is
>preferable to not generating any project file at all.
>
No, it's not, see below.
> Let's say a user
>builds Subversion and later installs Perl 5.8. If he then opens up visual
>studio to build the Perl bindings, isn't it be better for him to see the
>__SWIG_PERL__ project and get a sensible error message like 'Library
>perl56.lib not found'
>
And the user then bombards the dev@ lists with questions about why his
perl bindings don't build. I've seen too many of these kinds of
problems. That's why we have (or should have) flags: --enable-perl,
--enable-python, etc. They can either be off by default, or enabled iff
both SWIG and the scriptiong language are installed.
> when he tries to build it than for him to open the
>workspace and not see the project listed at all? In the first case the user
>at least has an error message to go on, in the second case he's likely to
>wonder if its possible to build the Perl bindings at all in Visual Studio.
>
Exactly my point, but with a different twist. If we know the bindings
won't build, why bother to generate the project?
--
Brane Čibej <brane_at_xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Aug 18 02:15:17 2003