Re: [Gretl-devel] @-based string substitution and 'foreign'
Am 10.02.2017 um 02:16 schrieb Allin Cottrell:
On Thu, 9 Feb 2017, oleg_komashko(a)ukr.net wrote:
> Gretl have a finite number of foreign interfaces May be, a decent
> solution is in making the use of '@' foreign language-specific (or,
> python-specific) With R it is a luxurous feature: it makes transfer of
> non-series data very convinient I'd prefer for changes to be specific
> and not to remove this possibility from foreign R.
OK, we won't make any precipitate changes, and it may make sense to
treat different "foreign" languages differently. (We already handle
python specially on account of its idiosyncratic treatment of white
space as syntactically significant.)
But could you elaborate on how string substitution is convenient in
relation to R? (Maybe we could recommend an alternative approach, or
maybe not.)
I'm leaning more and more towards the position that unconditional
substitution would either have to be banned in 'foreign' or should be
user configurable (as in "foreign language=R --strsubst=on").
Because I have verified that the "@" character also has its own meaning
in R:
https://stat.ethz.ch/R-manual/R-devel/library/methods/html/slot.html
So if a script author requests string substitution for a foreign block,
it would be her/his responsibility to avoid such ambiguities. It could
also be useful to have this choice on a block-per-block basis and not
for the entire script (as the above suggestion would ensure).
Don't know how much effort the implementation would imply.
thanks,
sven