Re: [dev-context] A few questions
Am 07.08.2009 um 14:33 schrieb Aditya Mahajan:
When I started using context, I was surprised that context did not
have anything for optional arguments. We could have something like
\definesinglearugment\foo[#1]#2{...}
to be equal to
\def\foo{\dosingleargument\dofoo}
\def\dofoo[#1]#2{...}
and
\definesingleempty\foo[#1]#2{...]
to be equal to
\def\foo{\dosingleempty\dofoo}
\def\dofoo[#1]#2{...}
and similar things for \definedoublearugment, \definetripleargument,
etc. I am not convinced that this is a good thing, but using
something like this in the core may result in a 10% reduction in the
code size and might kill a few dodos :)
It's now your turn to extend the macro to more optional arguments.
\def\define
{\doifnextoptionalelse\dodefine\nodefine}
\def\nodefine{\dodefine[0]}
\def\dodefine[#1]#2%
{\doifnextoptionalelse{\dododefine[#1]#2}{\nododefine[#1]#2}}
\def\dododefine[#1]#2[#3]#4%
{\definecomplexorsimple#2
\setvalue{simple\strippedcsname#2}{\getvalue{complex
\strippedcsname#2}[#3]}%
\ifcase0#1\relax
\setvalue{complex\strippedcsname#2}{#4}%
\setvalue{simple\strippedcsname #2}{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3##4{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3##4##5{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3##4##5##6{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3##4##5##6##6{#4}%
\or\setvalue{complex\strippedcsname#2}[##1]##2##3##4##5##6##7##8{#4}%
\or\setvalue{complex\strippedcsname#2}
[##1]##2##3##4##5##6##7##8##9{#4}%
\fi}
\def\nododefine[#1]#2#3%
{\ifx#2\undefined
\else
\showmessage{systems}{4}{\string#2}%
\fi
\ifcase0#1\def#2{#3}%
\or\def#2##1{#3}%
\or\def#2##1##2{#3}%
\or\def#2##1##2##3{#3}%
\or\def#2##1##2##3##4{#3}%
\or\def#2##1##2##3##4##5{#3}%
\or\def#2##1##2##3##4##5##6{#3}%
\or\def#2##1##2##3##4##5##6##7{#3}%
\or\def#2##1##2##3##4##5##6##7##8{#3}%
\or\def#2##1##2##3##4##5##6##7##8##9{#3}%
\else\def#2{#3}%
\fi}
\starttext
\define\foo{bar}
\define[2]\bar{bar:#1:#2}
\define[2]\help[me]{help #1 #2}
\foo
\bar{a}{b}
\help{john}
\help[you]{mike}
\stoptext
Wolfgang
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
Wolfgang Schuster wrote:
It's now your turn to extend the macro to more optional arguments.
hm, your challenging me? i must be crazy but here you go
\startluacode
local texsprint, ctxcatcodes = tex.sprint, tex.ctxcatcodes
local format, match, gmatch, rep = string.format, string.match,
string.gmatch, string.rep
local empty = {
single,
double,
triple,
quadruple,
quintuple,
}
local check = {
first,
second,
third,
fourth,
fifth,
}
function commands.define(str)
-- we could store the defaults in lua and call lua instead but
why bother
local arg, cmd = match(str,(.*)\\(.-)$)
local a = { }
for s in gmatch(arg,%[(.-)%]) do
a[#a+1] = s
end
local n = tonumber(a[#a])
if n then
a[#a] = nil
else
n = 0
end
texsprint(ctxcatcodes,format(\\unexpanded\\def\\%s,cmd))
if #a 0 then
texsprint(ctxcatcodes,format({\\do%sempty\\do%s},empty[#a],cmd))
texsprint(ctxcatcodes,format(\\def\\do%s,cmd))
for i=1,#a do
texsprint(ctxcatcodes,[#,i,])
end
texsprint(ctxcatcodes,{)
for i=#a,1,-1 do
texsprint(ctxcatcodes,format(\\if%sargument,check[i]))
texsprint(ctxcatcodes,format(\\def\\next{\\dodo%s,cmd))
for j=1,#a-i do
texsprint(ctxcatcodes,format([%s],a[j]))
end
for j=1,i do
texsprint(ctxcatcodes,format([#%s],j))
end
texsprint(ctxcatcodes,})
if i == 1 then
texsprint(ctxcatcodes,rep(\\fi,#a))
else
texsprint(ctxcatcodes,\\else)
end
end
texsprint(ctxcatcodes,\\next})
texsprint(ctxcatcodes,format(\\def\\dodo%s,cmd))
for i=1,#a do
texsprint(ctxcatcodes,[#,i,])
end
end
for i=1,n do
texsprint(ctxcatcodes,#,#a+i)
end
end
\stopluacode
\def\define#1#{\ctxlua{commands.define([[\detokenize{#1}]])}}
\starttext
\define[2]\whatevera{#1+#2}
\whatevera{A}{B}
\define[me][too][2]\whateverb{#1+#2+#3+#4}
\whateverb[A]{B}{C}
\whateverb[A][B]{C}{D}
\stoptext
we might add some \longs here and there
-
Hans Hagen | PRAGMA ADE
Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
| www.pragma-pod.nl
-
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
Am 06.08.2009 um 23:42 schrieb Hans Hagen:
1. texdefinition
I like start/stoptexdefinition because it makes the source for
certain macros more readable
than \def allows me but is there a way to use it for macros without
parameters?
this works
\starttexdefinition test {}
oeps
\stoptexdefinition
[\test]
not very nice and don't look pleasant to me
we can consider something like:
\starttexdefinition test optional,optional,argument,argument
oeps
\stoptexdefinition
which then becomes
\def\test[#1][#2]#3#4%
etc but i'm only willing to support that if we can stick to
reasonable definitions
or even
\starttexdefinition test optional:ns,argument:name
oeps
\stoptexdefinition
which then automatically makes you a \getvariable{ns}{key} as well
as \name
or maybe like xml:
\starttexdefinition test ns:optional,argument
oeps
\stoptexdefinition
with then makes an \getvariable{ns}{first} and so
needs a bit of thinking
i get the feeling we're moving towards latex3 (brr), before this
happens drop the whole
thing and i don't like \getvariable in this context anyway
2. commalists
Together with texdefinition I use now start/stopprocesscommalist
(feature request: I need also
start/stopprocessassignmentlist) but I can use the list value
without the need to define a command
where I specify what should happen with it
\startprocesscommalist[a,b,c,...]
the current list value is #1\par
\stopprocesscommalist
but why is there not something equivalent for the normal
processcommalist because the start/stop
method is not the best method when you need nested lists, something
like this would be nice then
(I know \doprocesscommalist is already defined):
\doprocesscommalist[a,b,c,...]%
{\processcommalist[x,y,z,...]%
{do something with #1 and ##1}}
hm, probably possible but not trivial (maybe not even robust) ...
needs thinking (also a bit slower as we explicitly need a trick
similar to the one used in dorecurse
when the inner list is used inside of a group the following works
\long\def\processcommalistwithcontent[#1]#2%
{\def\currentcommalistcommand##1{#2}%
\processcommalist[#1]\currentcommalistcommand}
\processcommalistwithcontent[a,b,c]
{{\processcommalistwithcontent[x,y,z]{This is subelement ##1 of
element #1.\par}}}
but the extra pair of braces is kind of akward. Working without them
is not the problem
\newcount\commalistdepth
\long\def\processcommalistwithcontent[#1]#2%
{\advance\commalistdepth\plusone
\long\expandafter\def\csname list:\number\commalistdepth
\endcsname##1{#2}%
\def\currentcommalistcommand{\csname list:\number\commalistdepth
\endcsname}%
\processcommalist[#1]\currentcommalistcommand
\advance\commalistdepth\minusone}
\processcommalistwithcontent[a,b,c]
{\processcommalistwithcontent[x,y,z]{This is subelement ##1 of element
#1.\par}}
but as you mentioned above some kind of check (depth counter) is needed.
3. setups
The various start/stopsetups environments another nice feature but
what's the correct method when
I want my own setup environment, should I adapt the following line
from core-env to my own command
\def\startsetups {\xxstartsetups\plusone \stopsetups }
\let\stopsetups \relax
and use method from lxml-ini to access/flush the content?
you mean that you want to pass an argument to a setup? (mandate in
xmlsetup)
yes, more or less what you do with xml in mkiv
\startxmlsetups ...
do this in that
\xmlflush{#1}
do more things
\stopxmlsetups
in my games module for the input for chess, go ...
\startpgnsetups move
do a few things before the move
\pgnflush{#1}
do a few things after the move
\stoppgnsetups
4. Mixed assignment/comma lists
Are mixed lists of normal comma elements and assignments are
allowed by ConTeXt’s syntax because
there is a command in the source (\processassignlist) to process
the non assignment values from
the list? On the other side these values generate a error message
on the terminal where one can read
there is a missing '=' in the element and this message can't be
disabled (without hacks).
yeah, a matter of choice ... we can drop the message if users don't
mind
we shouldn't end up with too many variants and personally i don't
like the mix (i.e. foo==yes can be used)
if the context way is to avoid comma values and assignment values in
the same list the message should better remain
Wolfgang
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
Am 07.08.2009 um 12:02 schrieb Hans Hagen:
Wolfgang Schuster wrote:
not very nice and don't look pleasant to me
sure, but a pleasant solution takes me more time -)
that's why i asked, i wasn't able to write a pleasant one myself
i get the feeling we're moving towards latex3 (brr), before this
happens drop the whole
thing and i don't like \getvariable in this context anyway
there was a presentation about this ltx3 low level syntax and i
definitely don't want to have something like that; it's more that in
context we have this system of [optional]{whatever} that we might
want to support
me neither, the latex3 system is too complicated and one has to
remember what n, N, x ... means etc.
there are a few interesting things like namespaces for module/macros
but that's not worth the effort they take
in my games module for the input for chess, go ...
\startpgnsetups move
do a few things before the move
\pgnflush{#1}
do a few things after the move
\stoppgnsetups
well, maybe something like
\starttexsetups move
do a few things before the move
\pgnflush{#1}
do a few things after the move
\stoptexsetups
as generic case?
i have to think about this but when i think about the current progress
on the module i won't come back to this before 2010 :)
Wolfgang
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
On Fri, 7 Aug 2009, Hans Hagen wrote:
i get the feeling we're moving towards latex3 (brr), before this happens
drop the whole
thing and i don't like \getvariable in this context anyway
there was a presentation about this ltx3 low level syntax and i definitely
don't want to have something like that; it's more that in context we have
this system of [optional]{whatever} that we might want to support
me neither, the latex3 system is too complicated and one has to remember
what n, N, x ... means etc.
indeed, and also kind of redundant as the name should denote the usage, i.e.
imagine that in e.g. lua we did myfunction_takes_num_and_num
i would not be surprised if it's also quite slow (esp when used in core code)
and core code is not used by users anyway
there are a few interesting things like namespaces for module/macros but
that's not worth the effort they take
indeed, but keep in mind that in context we seldom have other interfaces that
[] and {}, no optoinal * etc
One good thing with LaTeX 3 syntax is that something like
\DeclareDocumentCommand\foo { o o m } will be (from what I understand)
equivalent to
\def\foo
{\dodoubleempty\dofoo}
\def\dofoo[#1][#2]#3{...}
When I started using context, I was surprised that context did not have
anything for optional arguments. We could have something like
\definesinglearugment\foo[#1]#2{...}
to be equal to
\def\foo{\dosingleargument\dofoo}
\def\dofoo[#1]#2{...}
and
\definesingleempty\foo[#1]#2{...]
to be equal to
\def\foo{\dosingleempty\dofoo}
\def\dofoo[#1]#2{...}
and similar things for \definedoublearugment, \definetripleargument, etc.
I am not convinced that this is a good thing, but using something like
this in the core may result in a 10% reduction in the code size and might
kill a few dodos :)
Aditya
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
Aditya Mahajan wrote:
When I started using context, I was surprised that context did not have
anything for optional arguments. We could have something like
\definesinglearugment\foo[#1]#2{...}
to be equal to
\def\foo{\dosingleargument\dofoo}
\def\dofoo[#1]#2{...}
and
\definesingleempty\foo[#1]#2{...]
to be equal to
\def\foo{\dosingleempty\dofoo}
\def\dofoo[#1]#2{...}
and similar things for \definedoublearugment, \definetripleargument,
etc. I am not convinced that this is a good thing, but using something
like this in the core may result in a 10% reduction in the code size and
might kill a few dodos :)
there is \definecomplexorsimple
anyhow, it does not save commands and does not reduce the code size as
you still need the intermediates; it might increase the format
generation time with a few millisec
Hans
-
Hans Hagen | PRAGMA ADE
Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
| www.pragma-pod.nl
-
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
Re: [dev-context] A few questions
Wolfgang Schuster wrote:
Hi Hans,
here are a few questions which I want to ask since a few days.
1. texdefinition
I like start/stoptexdefinition because it makes the source for certain
macros more readable
than \def allows me but is there a way to use it for macros without
parameters?
this works
\starttexdefinition test {}
oeps
\stoptexdefinition
[\test]
we can consider something like:
\starttexdefinition test optional,optional,argument,argument
oeps
\stoptexdefinition
which then becomes
\def\test[#1][#2]#3#4%
etc but i'm only willing to support that if we can stick to reasonable
definitions
or even
\starttexdefinition test optional:ns,argument:name
oeps
\stoptexdefinition
which then automatically makes you a \getvariable{ns}{key} as well as \name
or maybe like xml:
\starttexdefinition test ns:optional,argument
oeps
\stoptexdefinition
with then makes an \getvariable{ns}{first} and so
needs a bit of thinking
2. commalists
Together with texdefinition I use now start/stopprocesscommalist
(feature request: I need also
start/stopprocessassignmentlist) but I can use the list value without
the need to define a command
where I specify what should happen with it
\startprocesscommalist[a,b,c,...]
the current list value is #1\par
\stopprocesscommalist
but why is there not something equivalent for the normal
processcommalist because the start/stop
method is not the best method when you need nested lists, something like
this would be nice then
(I know \doprocesscommalist is already defined):
\doprocesscommalist[a,b,c,...]%
{\processcommalist[x,y,z,...]%
{do something with #1 and ##1}}
hm, probably possible but not trivial (maybe not even robust) ... needs
thinking (also a bit slower as we explicitly need a trick similar to the
one used in dorecurse
3. setups
The various start/stopsetups environments another nice feature but
what's the correct method when
I want my own setup environment, should I adapt the following line from
core-env to my own command
\def\startsetups {\xxstartsetups\plusone \stopsetups }
\let\stopsetups \relax
and use method from lxml-ini to access/flush the content?
you mean that you want to pass an argument to a setup? (mandate in
xmlsetup)
4. Mixed assignment/comma lists
Are mixed lists of normal comma elements and assignments are allowed by
ConTeXt’s syntax because
there is a command in the source (\processassignlist) to process the non
assignment values from
the list? On the other side these values generate a error message on the
terminal where one can read
there is a missing '=' in the element and this message can't be disabled
(without hacks).
yeah, a matter of choice ... we can drop the message if users don't mind
we shouldn't end up with too many variants and personally i don't like
the mix (i.e. foo==yes can be used)
Hans
-
Hans Hagen | PRAGMA ADE
Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
| www.pragma-pod.nl
-
___
dev-context mailing list
dev-context@ntg.nl
http://www.ntg.nl/mailman/listinfo/dev-context
