[GenABEL-dev] [Genabel-commits] r1137 - tutorials/GenABEL_general

[Yurii Aulchenko]

> Great news! And congratulations with this milestone. The tutorial has
> always gotten a lot of positive comments. I'm sure that having it out in
> the open will make it even better.
>
>
This is the hope :) - and given there are already a couple of fixes (by you
:) ) things look good to me

>
> > For me, it works fine, e.g.
> >
> > svn export svn://svn.r-forge.r-project.org/svnroot/genabel/tutorials
> > <http://svn.r-forge.r-project.org/svnroot/genabel/tutorials>
> > cd tutorials/GenABEL_general
> > make
> >
> > The latter takes some minutes, but then I can open GenABEL-tutorial.pdf
> :)
>
> I ran into the following:
> - I didn't have MetABEL or MixABEL installed, so at some point the make
> run aborted. Would it make sense to let each chapter install the
> packages it needs? You could argue that something like that isn't needed
> because make is only to be run by developers who know how to do it. Or
> that you don't want to mess with the user's installed packages. On the
> other hand it would make the system more robust: only R, LaTeX, svn,
> ProbABEL and make would be the dependencies as far as I can see now.


I totally missed this problem. I think we should not mess up with the
user's set up, I would actually think whether we can make a kind of
'checks' (as they are in autoconf) so that if some package is not
installed, instead of the part tied to this package a page is generated
saying that the package is required...

I actually think this is among the smaller of the troubles :)



> Oh
> and MixABEL, which doesn't seem to be on CRAN yet.
>
>
This is true. I actually have no problem to (try to) publish it on CRAN,
but I am not the only person to have a say - we should definitely check
with William (William, please comment if you are reading this :) )

Next thing is that MixA requires GSL, so I am not sure how useful the CRAN
version - if they ever accept it - will be for a general user, I always had
a feeling that this is kind of 'advanced' skill to get MixA working.



>
> I'd say the Makefile looks good. Especially for someone who claims not
> to be good in Makefiles :-).


This is flattering :) I have copied it from someone else (the power of
community support!), and I do not quite understand how it works :)


>
> Another question popped up while looking at the files: what is your/our
> policy on the style of R that is in the tutorial. I see you abbreviate a
> lot of options (e.g. head=T instead of header=TRUE when using
> read.table()). Personally I always write these out in full because it
> reduces possible errors and it is enforced by CRAN I think (and Emacs
> makes the TRUEs and FALSEs light up nicely ;-)).
>
>
I agree that full argument names should be preferred. I was just being
lazy; I am trying to improve. If other people come up with things which are
"not in good style" we should probably think of a "good style rules"
document and then ask people to please to follow them.

best wishes,
Yurii
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20130314/098d1aeb/attachment.html>