References: <4DCA93CC.5020605@morningstar2.co.uk>
            <19920.15087.714131.100464@morse.mittelbach-online.de>
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="0-99627480-1305493460=:31292"
Message-ID:  <447088.31292.qm@web24704.mail.ird.yahoo.com>
Date:         Sun, 15 May 2011 22:04:20 +0100
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@listserv.uni-heidelberg.de>
Sender: Mailing list for the LaTeX3 project <LATEX-L@listserv.uni-heidelberg.de>
From: Chris A Rowley <ca_rowley@YAHOO.CO.UK>
Subject: Re: xparse and space skipping
To: LATEX-L@listserv.uni-heidelberg.de
In-Reply-To:  <19920.15087.714131.100464@morse.mittelbach-online.de>
Precedence: list
Status: R

--0-99627480-1305493460=:31292
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Another common use-case where (as with the AMS \\ in maths)=0Asomething non=
-standard seems intuitive is things=A0=0A'\item-like \foo s'=A0=0A=0A=0AWhe=
re=0A=0A\foo [arg] =A0 has the standard 'argument meaning'=0A=0Abut=0A=0A\f=
oo=0A[Paranthetical text starts here. ...=0A=0A=0Adoes not.=0A=0A=0A----- O=
riginal Message -----=0AFrom: Frank Mittelbach <frank.mittelbach@LATEX-PROJ=
ECT.ORG>=0ATo: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE=0ACc: =0ASent: Sunday, 15=
 May 2011, 21:43=0ASubject: Re: xparse and space skipping=0A=0AJoseph Wrigh=
t writes:=0A> You may remember that a while ago we discussed space skipping=
 for=0A> optional arguments with xparse, for example=0A> =0A>=A0 =A0 \foo{m=
andatory}[optional] other text=0A>=A0 =A0 \foo{mandatory} [optional] other =
text=0A>=A0 =A0 \foo{mandatory}=A0 other text=0A> =0A> At the time, the fee=
ling was that we should not skip spaces in these=0A> cases, so only the fir=
st of my examples is parsed as containing an=0A> optional argument.=0A=0AMu=
st confess that I can't remember that discussion nor the feeling that=0Ares=
ulted from it, but regardless of what I may have argued for in that=0Adiscu=
ssion (if I was part of it) looking at the above scheme I don't think I=0Al=
ike it or think it would be helpful.=0A=0AFor starters in my opinion xparse=
 is a module made to largely mimic the look=0Aand feel of current 2e docume=
nt level approach and what you describe here is a=0Astrong deviation from t=
hat. (this is independent of the fatc that there may be=0Aa completely diff=
erent approach to document level syntax, eventually, but this=0Ais not what=
 xparse was meant to be).=0A=0AIn a syntax of that sort, there needs to be =
good ways to naturally break up a=0Acommand and its arguments in different =
lines and a natural way to do that is=0Ato break between arguments (rather =
than inside) often even putting longer ones=0Aon single lines,eg=0A=0A=A0  =
\foobarbaz {aaaaaaaaaaaaaaaaaaaaaaaaa}=0A=A0 =A0 =A0 =A0 =A0 =A0 =A0 [bbbbb=
bbbbbbbbbbbbbbbbbbbb]=0A=A0 =A0 =A0 =A0 =A0 =A0 =A0 [ccc]=0A=0Aor even=0A=
=0A=A0  \foobarbaz=0A=A0 =A0  {aaaaaaaaaaaaaaaaaaaaaaaaa}=0A=A0 =A0  [bbbbb=
bbbbbbbbbbbbbbbbbbbb]=0A=A0 =A0  [ccc]=0A=0ATo be forced to keep the arg bo=
undaries on a single line or add %s all over=0Athe place is just not the wa=
y people currently use LaTeX (besides I think it=0Arather looks fairly ugly=
 and unreadable).=0A=0AAnd 2e is consistent here. They only well-known exce=
ption is \\ in the amsmath=0Areimplementation. To my knowledge this is real=
ly the only exception - if I'm=0Awrong please somebody tell me what else is=
 not following the general skipping=0Aof spaces scheme.=0A=0ANow one could =
argue that that this=A0 behavior for \\ is useful (especially in=0Amath for=
 which amsmath reimplements it) but realistically what are the other=0Aplac=
es that need this kind of behavior?=0A=0AThe problem is this type of behavi=
or is really most wanted in fairly simple=0Acommands, say with just one opt=
ional argument or just a single star plus=0Aperhaps optional argument. And =
there you run into the snag that you either=0Ahave to implement a full blow=
n scanner yourself and=A0 disable TeX's internal=0Ascanner completely (such=
 as turning anything into catcode 13 and assemble=0Aeverything yourself). O=
f course that is possible, but ...=0A=0AIf you don't want to go that far th=
en commands like \linebreak etc will always=0Asuffer from the fact that the=
re will be spaces allowed between them and a=0Afollowing [.=0A=0AAnd in fac=
t the general sentiment of this all needs to be configurable=0Aindividually=
 for any command is in my opinion a bad guide. It just mean that=0Anobody k=
nows what will happen on many commands as there is no rule other than=0Aloo=
king up how there are defined. And in my opinion this makes up an ugly user=
=0Ainterface. =0A=0Ahere is my summary=0A=0Aallow spaces but not \par betwe=
en arguments=0A=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=0A=0A=
pros=0A- consistent=0A- like 2e=0A- easy to remember=0A- makes for readable=
 sources on the whole=0A=0Acons=0A- would not support the \\ case as implem=
ented by amsmath=0A=0A=0Ado not allow spaces between arguments=0A=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=0A=0Apros=0A- consistent=0A- easy to remember=
=0A- would support the \\ case as implemented by amsmath=0A=0Acons=0A- comp=
letely different to 2e behavior=0A- makes for fairly unreadable sources on =
the whole=0A=0A=0Aleave it to the command how spaces between arguments are =
interpreted=0A=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=0A=0A=
pros=0A- would support the \\ case as implemented by amsmath=0A- makes for =
fairly unreadable sources on the whole=0A=0Acons=0A- inconsistent interface=
=0A- difficult to remember how individual commands behave=0A- more likely t=
o intorduce errors where you find argumnts ending up in text=0Aby mistake=
=0A=0A=0A> So looking again at this I think we should 'follow TeX', and be=
=0A> consistent in skipping spaces in all cases. I don't like the fact that=
=0A> at present there is a somewhat convoluted explanation of the behaviour=
=0A> in the xparse documentation: this tends to show up when something is n=
ot=0A> really correct! (At the same time, the implementation would be sligh=
tly=0A> easier to follow if this change was made.)=0A> =0A> Does this seem =
reasonable?=0A=0Ato me yes.=0A=0AThere would still be the question of contr=
ol symbols viz control words. By=0Adefault a control symbol (e.g., \? \/ ..=
.) will not skip spaces and \\=0Aactually explicitly has code to scan and i=
gnore spaces so was deliberately put=0Ainto the command class by Leslie to =
get a consistent interface for his main=0Acommands (and amsmath changed it =
back).=0A=0ABut I find a single exception (if implemented) still preferable=
 to the other=0Aoptions.=0A=0Afrank=0A
--0-99627480-1305493460=:31292
Content-Type: text/html; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

<html><body><div style=3D"color:; background-color:; font-family:times new =
roman, new york, times, serif;font-size:12pt"><div><span>Another common use=
-case where (as with the AMS \\ in maths)</span></div><div><span>something =
non-standard seems intuitive is things&nbsp;</span></div><div>'\item-like \=
foo s'&nbsp;<br></div><div><br></div><div>Where</div><div><br></div><div>\f=
oo [arg] &nbsp; has the standard 'argument meaning'</div><div><br></div><di=
v>but</div><div><br></div><div>\foo</div><div>[Paranthetical text starts he=
re. ...<br><div><br></div><div>does not.</div><div><br></div><div><br></div=
><div>----- Original Message -----<br>From: Frank Mittelbach &lt;frank.mitt=
elbach@LATEX-PROJECT.ORG&gt;<br>To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE<br>C=
c: <br>Sent: Sunday, 15 May 2011, 21:43<br>Subject: Re: xparse and space sk=
ipping<br><br>Joseph Wright writes:<br> &gt; You may remember that a while =
ago we discussed space skipping for<br> &gt; optional arguments with
 xparse, for example<br> &gt; <br> &gt;&nbsp; &nbsp; \foo{mandatory}[option=
al] other text<br> &gt;&nbsp; &nbsp; \foo{mandatory} [optional] other text<=
br> &gt;&nbsp; &nbsp; \foo{mandatory}&nbsp; other text<br> &gt; <br> &gt; A=
t the time, the feeling was that we should not skip spaces in these<br> &gt=
; cases, so only the first of my examples is parsed as containing an<br> &g=
t; optional argument.<br><br>Must confess that I can't remember that discus=
sion nor the feeling that<br>resulted from it, but regardless of what I may=
 have argued for in that<br>discussion (if I was part of it) looking at the=
 above scheme I don't think I<br>like it or think it would be helpful.<br><=
br>For starters in my opinion xparse is a module made to largely mimic the =
look<br>and feel of current 2e document level approach and what you describ=
e here is a<br>strong deviation from that. (this is independent of the fatc=
 that there may be<br>a completely different approach to document
 level syntax, eventually, but this<br>is not what xparse was meant to be).=
<br><br>In a syntax of that sort, there needs to be good ways to naturally =
break up a<br>command and its arguments in different lines and a natural wa=
y to do that is<br>to break between arguments (rather than inside) often ev=
en putting longer ones<br>on single lines,eg<br><br>&nbsp;  \foobarbaz {aaa=
aaaaaaaaaaaaaaaaaaaaaa}<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;=
 [bbbbbbbbbbbbbbbbbbbbbbbbb]<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &=
nbsp; [ccc]<br><br>or even<br><br>&nbsp;  \foobarbaz<br>&nbsp; &nbsp;  {aaa=
aaaaaaaaaaaaaaaaaaaaaa}<br>&nbsp; &nbsp;  [bbbbbbbbbbbbbbbbbbbbbbbbb]<br>&n=
bsp; &nbsp;  [ccc]<br><br>To be forced to keep the arg boundaries on a sing=
le line or add %s all over<br>the place is just not the way people currentl=
y use LaTeX (besides I think it<br>rather looks fairly ugly and unreadable)=
.<br><br>And 2e is consistent here. They only well-known exception
 is \\ in the amsmath<br>reimplementation. To my knowledge this is really t=
he only exception - if I'm<br>wrong please somebody tell me what else is no=
t following the general skipping<br>of spaces scheme.<br><br>Now one could =
argue that that this&nbsp; behavior for \\ is useful (especially in<br>math=
 for which amsmath reimplements it) but realistically what are the other<br=
>places that need this kind of behavior?<br><br>The problem is this type of=
 behavior is really most wanted in fairly simple<br>commands, say with just=
 one optional argument or just a single star plus<br>perhaps optional argum=
ent. And there you run into the snag that you either<br>have to implement a=
 full blown scanner yourself and&nbsp; disable TeX's internal<br>scanner co=
mpletely (such as turning anything into catcode 13 and assemble<br>everythi=
ng yourself). Of course that is possible, but ...<br><br>If you don't want =
to go that far then commands like \linebreak etc will
 always<br>suffer from the fact that there will be spaces allowed between t=
hem and a<br>following [.<br><br>And in fact the general sentiment of this =
all needs to be configurable<br>individually for any command is in my opini=
on a bad guide. It just mean that<br>nobody knows what will happen on many =
commands as there is no rule other than<br>looking up how there are defined=
. And in my opinion this makes up an ugly user<br>interface. <br><br>here i=
s my summary<br><br>allow spaces but not \par between arguments<br>=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D<br><br>pros<br> - consiste=
nt<br> - like 2e<br> - easy to remember<br> - makes for readable sources on=
 the whole<br><br>cons<br> - would not support the \\ case as implemented b=
y amsmath<br><br><br>do not allow spaces between arguments<br>=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D<br><br>pros<br> - consistent<br> - easy to remembe=
r<br> - would support the \\ case as implemented by amsmath<br><br>cons<br>=
 -
 completely different to 2e behavior<br> - makes for fairly unreadable sour=
ces on the whole<br><br><br>leave it to the command how spaces between argu=
ments are interpreted<br>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D<br><br>pros<br> - would support the \\ case as implemented by amsmat=
h<br> - makes for fairly unreadable sources on the whole<br><br>cons<br> - =
inconsistent interface<br> - difficult to remember how individual commands =
behave<br> - more likely to intorduce errors where you find argumnts ending=
 up in text<br>by mistake<br><br><br> &gt; So looking again at this I think=
 we should 'follow TeX', and be<br> &gt; consistent in skipping spaces in a=
ll cases. I don't like the fact that<br> &gt; at present there is a somewha=
t convoluted explanation of the behaviour<br> &gt; in the xparse documentat=
ion: this tends to show up when something is not<br> &gt; really correct! (=
At the same time, the implementation would be
 slightly<br> &gt; easier to follow if this change was made.)<br> &gt; <br>=
 &gt; Does this seem reasonable?<br><br>to me yes.<br><br>There would still=
 be the question of control symbols viz control words. By<br>default a cont=
rol symbol (e.g., \? \/ ...) will not skip spaces and \\<br>actually explic=
itly has code to scan and ignore spaces so was deliberately put<br>into the=
 command class by Leslie to get a consistent interface for his main<br>comm=
ands (and amsmath changed it back).<br><br>But I find a single exception (i=
f implemented) still preferable to the other<br>options.<br><br>frank<br></=
div></div><div style=3D"background-image: initial; background-attachment: i=
nitial; background-origin: initial; background-clip: initial; background-co=
lor: rgb(255, 255, 191); border-top-width: 0px; border-right-width: 0px; bo=
rder-bottom-width: 0px; border-left-width: 0px; border-style: initial; bord=
er-color: initial; color: rgb(0, 0, 0); margin-top: 0px; margin-right:
 0px; margin-bottom: 0px; margin-left: 0px; padding-top: 0px; padding-right=
: 0px; padding-bottom: 0px; padding-left: 0px; font-family: arial, sans-ser=
if; font-size: 13px; font-style: normal; font-variant: normal; font-weight:=
 normal; height: auto; line-height: normal; text-align: left; width: auto; =
direction: ltr; z-index: 99995; background-position: initial initial; backg=
round-repeat: initial initial; "></div></div></body></html>
--0-99627480-1305493460=:31292--