Hi, I'm going to try to ask this question without a minimal example for 
the moment, but if needed I'll make one.

I have a class file based on memoir that I'm working on, which has some 
conditional code based on whether a particular package is loaded:

\@ifpackageloaded{foo}{%
some code
}
{\relax}
}

This works fine in my class.  I thought I would replace <some code> in 
the class file with \input{file-containing-code} so that the fix is more 
modular. (It will not be needed when a new version the foo package is 
available).

But when I do this, I get a bunch of undefined control sequences, as if 
none of the memoir commands had been loaded.

Am I missing something here?  Is there any reason why using \input as 
opposed to having the code directly in the class file should work 
differently?

Thanks

Alan

In article <amunn-0B5168.15190317042010@news.eternal-september.org>,
 Alan Munn <amunn@msu.edu> wrote:

> Hi, I'm going to try to ask this question without a minimal example for 
> the moment, but if needed I'll make one.
> 
> I have a class file based on memoir that I'm working on, which has some 
> conditional code based on whether a particular package is loaded:
> 
> \@ifpackageloaded{foo}{%
> some code
> }
> {\relax}
> }

And, no, this extra } is not in my file. :-) 

> 
> This works fine in my class.  I thought I would replace <some code> in 
> the class file with \input{file-containing-code} so that the fix is more 
> modular. (It will not be needed when a new version the foo package is 
> available).
> 
> But when I do this, I get a bunch of undefined control sequences, as if 
> none of the memoir commands had been loaded.
> 
> Am I missing something here?  Is there any reason why using \input as 
> opposed to having the code directly in the class file should work 
> differently?
> 
> Thanks
> 
> Alan

Le 17/04/2010 21:19, Alan Munn a �crit :
> Hi, I'm going to try to ask this question without a minimal example for
> the moment, but if needed I'll make one.

Yes it could help. Or say the order of the operations :
\ProvideClass{myclass}

\LoadClass(memoir}

\@ifpackageloaded{foo}
	{\InputIfFileExists{...}}% or \usepackage ?
	\relax


>
> I have a class file based on memoir that I'm working on, which has some
> conditional code based on whether a particular package is loaded:
>
> \@ifpackageloaded{foo}{%
> some code
> }
> {\relax}
> }
>
> This works fine in my class.  I thought I would replace<some code>  in
> the class file with \input{file-containing-code} so that the fix is more
> modular. (It will not be needed when a new version the foo package is
> available).
>
> But when I do this, I get a bunch of undefined control sequences, as if
> none of the memoir commands had been loaded.
>
> Am I missing something here?  Is there any reason why using \input as
> opposed to having the code directly in the class file should work
> differently?
>
> Thanks
>
> Alan

In article <4bca0ab2$0$18408$426a74cc@news.free.fr>,
 GL <gouailles@gmail.com> wrote:

> Le 17/04/2010 21:19, Alan Munn a �crit :
> > Hi, I'm going to try to ask this question without a minimal example for
> > the moment, but if needed I'll make one.
> 
> Yes it could help. Or say the order of the operations:

Thanks.  I looked back at the code, and realise that the 
\@ifpackageloaded command was within the scope of \AtBeginDocument.  
Moving it outside of that scope fixed the problem.

Alan

Alan Munn <amunn@msu.edu> wrote:

> In article <4bca0ab2$0$18408$426a74cc@news.free.fr>,
>  GL <gouailles@gmail.com> wrote:
> 
> > Le 17/04/2010 21:19, Alan Munn a �crit :
> > > Hi, I'm going to try to ask this question without a minimal example for
> > > the moment, but if needed I'll make one.
> > 
> > Yes it could help. Or say the order of the operations:
> 
> Thanks.  I looked back at the code, and realise that the 
> \@ifpackageloaded command was within the scope of \AtBeginDocument.  
> Moving it outside of that scope fixed the problem.

Inside \AtBeginDocument you have probably the wrong catcode
for `@'.
  \AtBeginDocument{%
    \@ifpackageloaded{foo}{%
      \@ifpackagelater{foo}{2010/01/01}{% version of fixed foo
        \makeatletter
        \InputIfFileExists{fix-foo}{}{}%
        \makeatother
      }{}%
    }{}%
  }%

Or a little more defensive (retaining the catcode of @):

Instead of \makeatletter...\makeatother:

\begingroup
  \edef\x{\endgroup
    \noexpand\makeatletter
    \noexpand\InputIfFileExists{fix-foo}{}{}%
    \catcode64=\the\catcode64\relax % 64 = @
  }%
\x

-- 
Heiko Oberdiek

In article <hqdj57$dhp$1@news.eternal-september.org>,
 Heiko Oberdiek <heiko.oberdiek@googlemail.com> wrote:

> Alan Munn <amunn@msu.edu> wrote:
> 
> > In article <4bca0ab2$0$18408$426a74cc@news.free.fr>,
> >  GL <gouailles@gmail.com> wrote:
> > 
> > > Le 17/04/2010 21:19, Alan Munn a �crit :
> > > > Hi, I'm going to try to ask this question without a minimal example for
> > > > the moment, but if needed I'll make one.
> > > 
> > > Yes it could help. Or say the order of the operations:
> > 
> > Thanks.  I looked back at the code, and realise that the 
> > \@ifpackageloaded command was within the scope of \AtBeginDocument.  
> > Moving it outside of that scope fixed the problem.
> 
> Inside \AtBeginDocument you have probably the wrong catcode
> for `@'.
>   \AtBeginDocument{%
>     \@ifpackageloaded{foo}{%
>       \@ifpackagelater{foo}{2010/01/01}{% version of fixed foo
>         \makeatletter
>         \InputIfFileExists{fix-foo}{}{}%
>         \makeatother
>       }{}%
>     }{}%
>   }%
> 
> Or a little more defensive (retaining the catcode of @):
> 
> Instead of \makeatletter...\makeatother:
> 
> \begingroup
>   \edef\x{\endgroup
>     \noexpand\makeatletter
>     \noexpand\InputIfFileExists{fix-foo}{}{}%
>     \catcode64=\the\catcode64\relax % 64 = @
>   }%
> \x

Thanks, Heiko.  That definitely was the problem, and now I understand 
why.

Alan