On Thu, 4 Jul 2019, Sven Schreiber wrote:
Am 02.07.2019 um 08:22 schrieb Riccardo (Jack) Lucchetti:
> That said, I'll have a go at cleaning all my funtion packages from
> isnull() and isstring() at the earliest opportunity, and I would advise
> everyone to do the same.
I think this call is buried too deep in this thread, we have to be
"louder" in a sense.
Agreed, if we really want to get rid of isnull().
Allin, you already checked (via grep, presumably) which packages
have
isnull() in them. Could you provide the list, and then we can post that,
perhaps on the user list.
No current packages or addons use isstring(). Here are those using
isnull():
a_eff
arima_sim
armax
BACE
cmatrix
delayspectral
fcModels
FEP
FEPoisson
getQuandl
heckprobit
ivintreg
lagreg
margeff_el
matrix_perf
poisson_mfx
roc
sfa_mod
tobit_mfx
tsfcst
waldTest
However (despite my premature attempt to purge isnull!) I'm OK with
being fairly relaxed about this. Right now "isnull" is no longer a
function in its own right (and its documentation is gone); it's an
alias to exists() which reverses the sense of the latter.
I don't really see any pressing reason to bother package writers
about this, if we're willing to tolerate the alias. I think it would
be sufficient to say something like, "If you have any independent
reason to update your package, please substitute !exists() for
isnull() while you're at it."
And next time (if ever) that I'm tempted to trash "isnull" I'll take
a proper look at all packages and addons first!
Allin