5:12 p.m.
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to
"good old" printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If
lots of people tell me they still think printf sucks, than I guess
I'll have to rethink.
<hansl>
set echo off
set messages off
# hansl, using printf
h = 8
w = 6
printf "Henry the %gth had %g wives\n", h, w
vname = "dispinc"
b = 0.9307
se = 0.0421
printf "The coefficient on %s is %g, with std error %g\n",
vname, b, se
foreign language=R
h <- 8
w <- 6
cat(paste("Henry the", h, "th had", w, "wives\n"))
vname <- "dispinc"
b <- 0.9307
se <- 0.0421
cat(paste("The coefficient on", vname, "is", b, ", with std
error",
se, "\n"))
end foreign
foreign language=Ox
#include <oxstd.h>
main() {
decl h = 8;
decl w = 6;
print("Henry the ", h, "th had ", w, " wives\n");
decl vname = "dispinc";
decl b = 0.9307;
decl se = 0.0421;
print("The coefficient on ", vname, " is ", b, " with std
error ",
se, "\n");
}
end foreign
</hansl>
Allin
5:32 p.m.
Of those, I personally like hansl & printf much better. It's taken me a while to
get used to (s)printf and it still occasionally gives me the irrits, but the more I play
around with it the easier it gets.
Fwiw,
PS
-----Original Message-----
From: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] On Behalf Of Allin Cottrell
Sent: Monday, September 19, 2016 5:12 PM
To: Gretl development
Subject: [Gretl-devel] printf poll?
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to "good
old"
printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If lots of
people tell me they still think printf sucks, than I guess I'll have to rethink.
<hansl>
set echo off
set messages off
# hansl, using printf
h = 8
w = 6
printf "Henry the %gth had %g wives\n", h, w vname = "dispinc"
b = 0.9307
se = 0.0421
printf "The coefficient on %s is %g, with std error %g\n",
vname, b, se
foreign language=R
h <- 8
w <- 6
cat(paste("Henry the", h, "th had", w, "wives\n")) vname
<- "dispinc"
b <- 0.9307
se <- 0.0421
cat(paste("The coefficient on", vname, "is", b, ", with std
error",
se, "\n"))
end foreign
foreign language=Ox
#include <oxstd.h>
main() {
decl h = 8;
decl w = 6;
print("Henry the ", h, "th had ", w, " wives\n");
decl vname = "dispinc";
decl b = 0.9307;
decl se = 0.0421;
print("The coefficient on ", vname, " is ", b, " with std
error ",
se, "\n");
}
end foreign
</hansl>
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
5:45 p.m.
I'll take the gretl version as well.
Lee
On Mon, Sep 19, 2016 at 4:12 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to
"good old" printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If lots of
people tell me they still think printf sucks, than I guess I'll have to
rethink.
<hansl>
set echo off
set messages off
# hansl, using printf
h = 8
w = 6
printf "Henry the %gth had %g wives\n", h, w
vname = "dispinc"
b = 0.9307
se = 0.0421
printf "The coefficient on %s is %g, with std error %g\n",
vname, b, se
foreign language=R
h <- 8
w <- 6
cat(paste("Henry the", h, "th had", w, "wives\n"))
vname <- "dispinc"
b <- 0.9307
se <- 0.0421
cat(paste("The coefficient on", vname, "is", b, ", with std
error",
se, "\n"))
end foreign
foreign language=Ox
#include <oxstd.h>
main() {
decl h = 8;
decl w = 6;
print("Henry the ", h, "th had ", w, " wives\n");
decl vname = "dispinc";
decl b = 0.9307;
decl se = 0.0421;
print("The coefficient on ", vname, " is ", b, " with std
error ",
se, "\n");
}
end foreign
</hansl>
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Lee Adkins
Professor of Economics and Department Head
Economics and Legal Studies in Business
Oklahoma State University
Email: lee.adkins(a)okstate.edu
Fax: 405-744-5180
Phone: 405-744-5196
URL: www.learneconometrics.com
6:05 p.m.
I prefer the Ox print() function, but printf would need to stick around for backward
compatibility. Moreover, it seems formatting using the Ox print() syntax would be more
complicated than printf. Thus, I am not sure it makes enough difference to warrant the
extra work.
Logan
From: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-bounces@lists.wfu.edu] On
Behalf Of Lee Adkins
Sent: Monday, September 19, 2016 4:46 PM
To: Gretl development <gretl-devel(a)lists.wfu.edu>
Subject: Re: [Gretl-devel] printf poll?
I'll take the gretl version as well.
Lee
On Mon, Sep 19, 2016 at 4:12 PM, Allin Cottrell
<cottrell@wfu.edu<mailto:cottrell@wfu.edu>> wrote:
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to
"good old" printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If lots of people tell
me they still think printf sucks, than I guess I'll have to rethink.
<hansl>
set echo off
set messages off
# hansl, using printf
h = 8
w = 6
printf "Henry the %gth had %g wives\n", h, w
vname = "dispinc"
b = 0.9307
se = 0.0421
printf "The coefficient on %s is %g, with std error %g\n",
vname, b, se
foreign language=R
h <- 8
w <- 6
cat(paste("Henry the", h, "th had", w, "wives\n"))
vname <- "dispinc"
b <- 0.9307
se <- 0.0421
cat(paste("The coefficient on", vname, "is", b, ", with std
error",
se, "\n"))
end foreign
foreign language=Ox
#include <oxstd.h>
main() {
decl h = 8;
decl w = 6;
print("Henry the ", h, "th had ", w, " wives\n");
decl vname = "dispinc";
decl b = 0.9307;
decl se = 0.0421;
print("The coefficient on ", vname, " is ", b, " with std error
",
se, "\n");
}
end foreign
</hansl>
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel@lists.wfu.edu<mailto:Gretl-devel@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Lee Adkins
Professor of Economics and Department Head
Economics and Legal Studies in Business
Oklahoma State University
Email: lee.adkins@okstate.edu<mailto:lee.adkins@okstate.edu>
Fax: 405-744-5180
Phone: 405-744-5196
URL: www.learneconometrics.com<http://www.learneconometrics.com>
6:06 p.m.
I also favor the gretl/hansl way which is similar to matlabs fprintf()
Artur
Am 19. September 2016 23:45:49 MESZ, schrieb Lee Adkins <lee.adkins(a)okstate.edu>:
I'll take the gretl version as well.
Lee
On Mon, Sep 19, 2016 at 4:12 PM, Allin Cottrell <cottrell(a)wfu.edu>
wrote:
> Coming off Sven's posting at
> http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
> I'd like to show a couple of real-world examples of alternatives to
> "good old" printf in the econometrics domain.
>
> I'd be interested to hear what people think of the comparison. If
lots of
> people tell me they still think printf sucks, than I guess I'll have
to
> rethink.
>
> <hansl>
> set echo off
> set messages off
>
> # hansl, using printf
> h = 8
> w = 6
> printf "Henry the %gth had %g wives\n", h, w
> vname = "dispinc"
> b = 0.9307
> se = 0.0421
> printf "The coefficient on %s is %g, with std error %g\n",
> vname, b, se
>
> foreign language=R
> h <- 8
> w <- 6
> cat(paste("Henry the", h, "th had", w, "wives\n"))
> vname <- "dispinc"
> b <- 0.9307
> se <- 0.0421
> cat(paste("The coefficient on", vname, "is", b, ", with std
error",
> se, "\n"))
> end foreign
>
> foreign language=Ox
> #include <oxstd.h>
> main() {
> decl h = 8;
> decl w = 6;
> print("Henry the ", h, "th had ", w, " wives\n");
> decl vname = "dispinc";
> decl b = 0.9307;
> decl se = 0.0421;
> print("The coefficient on ", vname, " is ", b, " with std
error ",
> se, "\n");
> }
> end foreign
> </hansl>
>
> Allin
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
--
Lee Adkins
Professor of Economics and Department Head
Economics and Legal Studies in Business
Oklahoma State University
Email: lee.adkins(a)okstate.edu
Fax: 405-744-5180
Phone: 405-744-5196
URL: www.learneconometrics.com
------------------------------------------------------------------------
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
1 a.m.
Nothing wrong with printf.
Keep it.
I don't feel any need for change.
Berend
On 19 Sep 2016, at 23:12, Allin Cottrell <cottrell(a)wfu.edu>
wrote:
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to
"good old" printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If lots of people tell
me they still think printf sucks, than I guess I'll have to rethink.
<hansl>
set echo off
set messages off
# hansl, using printf
h = 8
w = 6
printf "Henry the %gth had %g wives\n", h, w
vname = "dispinc"
b = 0.9307
se = 0.0421
printf "The coefficient on %s is %g, with std error %g\n",
vname, b, se
foreign language=R
h <- 8
w <- 6
cat(paste("Henry the", h, "th had", w, "wives\n"))
vname <- "dispinc"
b <- 0.9307
se <- 0.0421
cat(paste("The coefficient on", vname, "is", b, ", with std
error",
se, "\n"))
end foreign
foreign language=Ox
#include <oxstd.h>
main() {
decl h = 8;
decl w = 6;
print("Henry the ", h, "th had ", w, " wives\n");
decl vname = "dispinc";
decl b = 0.9307;
decl se = 0.0421;
print("The coefficient on ", vname, " is ", b, " with std error
",
se, "\n");
}
end foreign
</hansl>
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
1:37 a.m.
On Mon, 19 Sep 2016, Allin Cottrell wrote:
Coming off Sven's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2016-September/006903.html
I'd like to show a couple of real-world examples of alternatives to
"good old" printf in the econometrics domain.
I'd be interested to hear what people think of the comparison. If lots of
people tell me they still think printf sucks, than I guess I'll have to
rethink.
I'm a huge fan of (s)printf, so keeping it or not is not a meaningful
question IMO. Of course it has to stay, if only in the interest of
backward compatibility.
The issue here is: is it desirable to have a "dumbed down" version of
printf? As I said, the only case I can see some scope for is a
multi-argument print() _function_, pretty much ox-style (but without the
full Ox format syntax, which is even more confusing than printf,
methinks). I wouldn't use it very much myself, but I can see that some
people might like it. Possibly, a sister function println() could provide
automatic end-of-line at the end, Pascal-style (ok, that would be
write/writeln, but you get the point).
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
4:02 a.m.
Am 20.09.2016 um 07:37 schrieb Riccardo (Jack) Lucchetti:
On Mon, 19 Sep 2016, Allin Cottrell wrote:
>
> I'd be interested to hear what people think of the comparison. If lots
> of people tell me they still think printf sucks, than I guess I'll
> have to rethink.
>
I'm a huge fan of (s)printf, so keeping it or not is not a meaningful
question IMO. Of course it has to stay, if only in the interest of
backward compatibility.
The issue here is: is it desirable to have a "dumbed down" version of
printf?
Exactly. It seems to me everybody in this poll was answering the
question "Should printf() be replaced?". Which was never my intention.
The point is whether or not to *add* a simpler version for simple stuff.
thanks,
sven
4:33 p.m.
On Tue, 20 Sep 2016, Sven Schreiber wrote:
Am 20.09.2016 um 07:37 schrieb Riccardo (Jack) Lucchetti:
> On Mon, 19 Sep 2016, Allin Cottrell wrote:
>> I'd be interested to hear what people think of the comparison. If lots
>> of people tell me they still think printf sucks, than I guess I'll
>> have to rethink.
>
> I'm a huge fan of (s)printf, so keeping it or not is not a meaningful
> question IMO. Of course it has to stay, if only in the interest of
> backward compatibility.
>
> The issue here is: is it desirable to have a "dumbed down" version of
> printf?
Exactly. It seems to me everybody in this poll was answering the question
"Should printf() be replaced?". Which was never my intention.
The point is whether or not to *add* a simpler version for simple stuff.
I agree, and sorry I wasn't clearer. The idea behind my "poll" was
not to ask, Should we replace printf with something else? but
rather, Do we need to provide an alternative to printf for formatted
output?
We may not be able to reach consensus on this, but let me try a
little further. At present we have three commands in this
neighbourhood: print, printf and eval. Some might say that's already
too many (though actually I wouldn't agree, for reasons I've tried
to explain up-thread). However, surely it ought to be enough! And in
particular, I really don't want to introduce into hansl a
"competitor" to printf -- because, as I've already said, I think
that printf is the better design for complex mixtures of text and
numerics (though of course everyone doesn't have to agree with me on
that).
That said, I can see a case for making it easier to output very
simple stuff such as a string followed by a scalar on the same line.
In git there's an experiment to this effect; here's an example:
<hansl>
set echo off
set messages off
answer = 42
print "The number of roads a man must walk down is " -n
eval answer
</hansl>
The option -n (long form "--no-break") tells "print" not to add a
line-break; therefore the "eval" output follows directly:
<output>
The number of roads a man must walk down is 42
</output>
I don't know if anyone will consider this worthwhile. Certainly, one
can imagine a command that would produce the same effect with a
single line of input -- but, trust me, that would have to a new,
fourth, printing command; it could not be "print," on pain of
throwing away its facility to print the values of as many variables
as you like, given just their space-separated names.
Allin
1:31 a.m.
On Tue, 20 Sep 2016, Allin Cottrell wrote:
At present we have three commands in this neighbourhood: print,
printf and eval. Some might say that's already too many (though actually I
wouldn't agree, for reasons I've tried to explain up-thread). However, surely
it ought to be enough! And in particular, I really don't want to introduce
into hansl a "competitor" to printf -- because, as I've already said, I
think
that printf is the better design for complex mixtures of text and numerics
(though of course everyone doesn't have to agree with me on that).
That said, I can see a case for making it easier to output very simple stuff
such as a string followed by a scalar on the same line.
In git there's an experiment to this effect; here's an example:
<hansl>
set echo off
set messages off
answer = 42
print "The number of roads a man must walk down is " -n
eval answer
</hansl>
The option -n (long form "--no-break") tells "print" not to add a
line-break;
therefore the "eval" output follows directly:
<output>
The number of roads a man must walk down is 42
</output>
I don't know if anyone will consider this worthwhile.
I, for one, do not. And by the way, someone may see it as a whimsical and
geekish Unix-ism :) (compare the -n option to the "echo" shell command).
After having heard several opinions (thanks to everybody who joined the
debate) I suggest to leave everything as is now. The recent improvements
in the documentation should also help.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
5:41 a.m.
El 21/09/16 a las 07:31, Riccardo (Jack) Lucchetti escribió:
After having heard several opinions (thanks to everybody who joined
the debate) I suggest to leave everything as is now. The recent
improvements in the documentation should also help.
I also vote for that.
--
Ignacio Díaz-Emparanza
Departamento de Economía Aplicada III (Econometría y Estadística)
Universidad del País Vasco - Euskalherriko Unibertsitatea, UPV/EHU
Tfno: (+34) 94 601 3732
http://www.ehu.eus/ea3
7:46 a.m.
Am 21.09.2016 um 07:31 schrieb Riccardo (Jack) Lucchetti:
On Tue, 20 Sep 2016, Allin Cottrell wrote:
> That said, I can see a case for making it easier to output very
simple
> stuff such as a string followed by a scalar on the same line.
Yes, I was especially fascinated by Lee's Stata example:
di "The sample size is " e(N)
I'm ready to surrender and declare defeat, but not without a final note
that I find it sad that this level of simplicity shouldn't be possible
in hansl...
> answer = 42
> print "The number of roads a man must walk down is " -n
> eval answer
...
> I don't know if anyone will consider this worthwhile.
I, for one, do not.
Here I agree with Jack, this doesn't really cut it. However, it prompted
me to try this, which also produced a single-line output:
<hansl>
set echo off
printf "heyhey: "
eval 3/2
</hansl>
After having heard several opinions (thanks to everybody who joined the
debate) I suggest to leave everything as is now. The recent improvements
in the documentation should also help.
What I learned from the discussion is that I still think hansl could use
a syntactic-sugar/one-argument version of sprintf() such that sprintf(x)
== sprintf("%g", x), when x is a scalar type. (Could also have a new
alias name instead of 'sprintf' then, but that's not my point.)
thanks,
sven
4:49 p.m.
On Wed, 21 Sep 2016, Sven Schreiber wrote:
Am 21.09.2016 um 07:31 schrieb Riccardo (Jack) Lucchetti:
> On Tue, 20 Sep 2016, Allin Cottrell wrote:
>> That said, I can see a case for making it easier to output very simple
>> stuff such as a string followed by a scalar on the same line.
Yes, I was especially fascinated by Lee's Stata example:
di "The sample size is " e(N)
I'm ready to surrender and declare defeat, but not without a final
note that I find it sad that this level of simplicity shouldn't be
possible in hansl...
IMO, the simplicity is possible in hansl, for anyone allergic to
printf; it just takes two lines:
print "Comment on the following number:"
print number
(or "eval number" if it's an expression rather a variable)
What I learned from the discussion is that I still think hansl
could use a syntactic-sugar/one-argument version of sprintf() such
that sprintf(x) == sprintf("%g", x), when x is a scalar type.
(Could also have a new alias name instead of 'sprintf' then, but
that's not my point.)
That could be implemented without prejudice to other points if we
reckon it's worthwhile. I suppose the intended context would be
something like the following (please correct me if I don't have this
right):
<hansl>
r = 42
# variant 1
eval "Here's a number of interest: " ~ sprintf(r)
# versus variant 2
printf "Here's a number of interest: %g\n", r
</hansl>
OK, no more polls for now, but personally I doubt that many people
would find variant 1 simpler/more intuitive than variant 2.
Allin
8:07 a.m.
Am 21.09.2016 um 22:49 schrieb Allin Cottrell:
On Wed, 21 Sep 2016, Sven Schreiber wrote:
IMO, the simplicity is possible in hansl, for anyone allergic to
printf;
it just takes two lines:
You're right, the level of simplicity is comparable. I guess I
implicitly also meant "short" when I said simple. I am *not* allergic to
printf (using it all the time), but I guess I'm allergic to
two-lines-vs.-one, both in input and output. Clearly a personal taste,
apparently not shared by everyone.
That could be implemented without prejudice to other points if we
reckon
it's worthwhile. I suppose the intended context would be something like
the following (please correct me if I don't have this right):
It's an intended context, but I would also say it's useful/practical for
everyday assignment usage, as in:
string s = sprintf(r)
As I said, it's about a sensible default; having to always type the
additional five characters <"%g",> strikes me as an unnecessary
nuisance. This would also be a kind of inverse to the atof() function.
OK, no more polls for now,
Yes, no more polls, and thanks for having a constructive discussion.
Let me repeat that I was trying to take the perspective of somebody who
is not currently using gretl. While we all are enthusiastic about gretl,
I think we need to remember that there's lots of competition out there,
especially for hansl the language (perhaps less for gretl, the GUI
program). So I have been trying to point out what I perceived as
disadvantages of hansl relative to other languages.
thanks,
sven
3:25 p.m.
On Wed, 21 Sep 2016, Riccardo (Jack) Lucchetti wrote:
On Tue, 20 Sep 2016, Allin Cottrell wrote:
> [...] I can see a case for making it easier to output very simple
> stuff such as a string followed by a scalar on the same line. In
> git there's an experiment to this effect; here's an example:
>
> <hansl>
> set echo off
> set messages off
>
> answer = 42
> print "The number of roads a man must walk down is " -n
> eval answer
> </hansl>
>
> The option -n (long form "--no-break") tells "print" not to add a
> line-break; therefore the "eval" output follows directly:
>
> <output>
> The number of roads a man must walk down is 42
> </output>
>
> I don't know if anyone will consider this worthwhile.
I, for one, do not. And by the way, someone may see it as a whimsical and
geekish Unix-ism :) (compare the -n option to the "echo" shell command).
Dang, I've been rumbled! OK, scratch that idea.
After having heard several opinions (thanks to everybody who joined
the
debate) I suggest to leave everything as is now. The recent improvements in
the documentation should also help.
I'd also like to thank everyone who weighed in on this. We'll go
with the status quo for the present -- with the option of revisiting
the issue later. I do think that the doc updates may help.
Allin
9:51 a.m.
Personally, I find printf very nice.
The users without c background may need
a self-sufficient tutorial, especially on different
numerical formats
Oleh
20 вересня 2016, 00:12:31, від "Allin Cottrell" <cottrell(a)wfu.edu>:
...
5:03 p.m.
On Tue, 20 Sep 2016, oleg_komashko(a)ukr.net wrote:
Personally, I find printf very nice.
The users without c background may need
a self-sufficient tutorial, especially on different
numerical formats
I've recently revised the discussion of printf in the hansl primer;
hopefully this may go some way towards answering your point.
Allin
2978
days inactive
2981
days old
16 comments
10 participants
participants (10)
-
Allin Cottrell -
Artur Tarassow -
Berend Hasselman -
Ignacio Diaz-Emparanza -
Lee Adkins -
Logan Kelly -
oleg_komashko＠ukr.net -
Riccardo (Jack) Lucchetti -
Summers, Peter -
Sven Schreiber