Path: csiph.com!usenet.pasdenom.info!weretis.net!feeder1.news.weretis.net!feeder.erje.net!newsfeed.xs4all.nl!newsfeed5.news.xs4all.nl!xs4all!post.news.xs4all.nl!not-for-mail
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:X-Mailer:References:Message-ID:Date:From:Reply-To:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=BcYa32P34I1ruLGyKkOSI2gf2/wjTOSnh/aEdxveuSp8eGzKWWyR/y4ODqfEjr/gl4yPqC4+Ne96YtoHQc65JEJT/7FRdlwh78ZgjJrFHbMBkM3de9MPwO9hoGIN74g+3KMwS7SF4PiSc4n0lPLeD6PaLTNqm3IqVDEg/Y8Oc7U=;
References: <CABgq=FyhR+Ldujj3YKRBpVXRVoeoayuXZviUUcNx-gXnFiHLSw@mail.gmail.com> <CACoeR0xV=28j6rFHm1ep3gzH9C0sbDqyxRRwX-JrVm3G+rrXww@mail.gmail.com> <1333174946.18436.YahooMailNeo@web121506.mail.ne1.yahoo.com> <jl6ma0$ptg$1@dough.gmane.org> <mailman.1200.1333352587.3037.python-list@python.org> <4f7962b0$0$29981$c3e8da3$5496439d@news.astraweb.com> <1333357906.6147.YahooMailNeo@web121503.mail.ne1.yahoo.com> <CAPTjJmqtMhS8jp3D+SNiOAp-v9QzP5tkQzCqmO_Re7oLepOWwQ@mail.gmail.com> <1333367201.47465.YahooMailNeo@web121506.mail.ne1.yahoo.com> <CAPTjJmoH9uCOTQ2M6fhQjUnydAGGBhhFo+++2i9HbUEFg5KJ2g@mail.gmail.com> <1333381766.14038.YahooMailNeo@web121504.mail.ne1.yahoo.com> <CAPTjJmrNFxH+d-X7TZuRttmTziOLv56s+3eE1HHtEEx4ARj9FA@mail.gmail.com>
Date: Mon, 2 Apr 2012 18:57:22 -0700 (PDT)
From: Yingjie Lan <lanyjie@yahoo.com>
Subject: Re: string interpolation for python
To: "python-list@python.org" <python-list@python.org>
In-Reply-To: <CAPTjJmrNFxH+d-X7TZuRttmTziOLv56s+3eE1HHtEEx4ARj9FA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable
Precedence: list
Reply-To: Yingjie Lan <lanyjie@yahoo.com>
Newsgroups: comp.lang.python
Message-ID: <mailman.1251.1333418251.3037.python-list@python.org>
Lines: 70
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:22565

>>>  I can't find a way to use an argument more than once,=0A=0A>>>  withou=
t switching to "dictionary mode" and using keys for =0A>>> everything.=0A=
=0A=0AEven in "dictionary mode", the key is spelled more than=0A=0Aonce. Th=
e "tuple mode" below seems to save some typing.=A0=0A=0AHowever, when there=
 are more and more items to format,=A0=0Athe readability=A0deteriorates qui=
ckly in the "tuple" mode.=0A=0AUse an argument more than once is, after all=
, probably a=0Anot-so-often use case.=0A=0A=0AThe "dictionary mode" greatly=
 enhances readability, but=A0=0Ayou have=A0to provide a dict object and mak=
e sure the name in=A0=0Ayour=A0formatting string matches=A0the keys in the =
dictionary.=0AAnd this matching requirement is because of information=0Ared=
undancy in the arrangement. And information redundancy=0Ais often the root =
of evil for various kinds of trouble.=0A=0ADynamic string seems to have the=
 good without redundancy.=0AAnd there is no need to build a dict for it.=A0=
=0AHowever, when there is already the dict for use,=A0=0Aclearly the dict f=
ormat is the winner.=0A>>  Ack.=0A>> =0A>>  In this case, you can use forma=
t:=0A>> =0A>>>>>  "Hello {0}, what's up with {1}? Hey, {0} I'm =0A> speakin=
g to you!".format=0A>>  ('George', 'Melissa')=0A>>  "Hello George, what's u=
p with Melissa? Hey, George I'm =0A> speaking to you!"=0A> =0A> Or don't bo=
ther with the initial string, and simply pass everything as=0A> arguments:=
=0A> =0A> def yingjie_format(*args):=0A> =A0 =A0 it=3Diter(args)=0A> =A0 =
=A0 return ''.join(s%next(it,None) for s in it)=0A> yingjie_format("sin(%s"=
,x,") =3D %0.3f", sin(x))=0A>=A0=0A=0AThat's very nice, thanks!=0A=0A> So i=
f they're exactly like normal expressions, why not simply use=0A=0A> normal=
 expressions?=0A=0A=0AI think use dynamic strings can have these benefits:=
=0A1) you less keys (punch less keys).=0A2) better readability (less clutte=
rs)=0A3) you don't have to explicit convert/format expressions into strings=
=0A4) better performance too (adding strings together is generally slow).=
=0A=A0=0A=0A>>>  sprintf("UPDATE tablename SET modified=3Dnow()%{,%s=3D:%[0=
]s%} WHERE=0A>>>  key=3D%d",array_of_field_names,primary_key_value)=0A>>>  =
--> "UPDATE tablename SET =0A> modified=3Dnow(),foo=3D:foo,bar=3D:bar,quux=
=3D:quux=0A>>>  WHERE key=3D1234"=0A>>> =0A>>>  You're still paying for no =
complexity you aren't actually =0A> using.=0A>>>  It's=A0clear and readable=
.=0A>> =0A>>  You are really good at that. Maybe not everybody is as=0A>>  =
experience as you, and I suppose the learning curve is=0A>>  kind of hard t=
o climb.=0A> =0A> Yes, it takes some learning to use it. But that's true of=
 everything,=0A> no less of your magical string interpolation. My point is =
that simple=0A> examples should be (and are, with printf formatting) simple=
, such that=0A> you only get those more complicated format strings when you=
're=0A> actually doing a complicated job (in that case, taking each element=
 of=0A> an array and using it twice - actually, it was taking the indices o=
f a=0A> mapping that would end up being passed to the DB query function, th=
us=0A> providing values to the :foo :bar variables).=0A>=A0=0A=0ASure. But =
if one thing does well on both simple and complex=A0=0Asituations, why not =
that thing?=0A=0A>>  Those expressions are embedded, you don't need eval()=
=0A>>  to have the result though. Are we on the same page?=0A> =0A> I can s=
ee three implementation paths:=0A> =0A> 1) Language feature. It really *is*=
 just an expression. There's no way=0A> that a user can provide them, so th=
ere's actually no similarity to=0A> eval. But this requires that Python its=
elf handle things.=0A> =0A> 2) Precompiler. It *becomes* an expression. Aga=
in, perfectly safe,=0A> although I don't know how useful this really is.=0A=
> =0A> 3) Functoin. As several have suggested, you could do some magic and=
=0A> use d("this is a $dollar$ $interpolated$ string") to implement. For=0A=
> this, you *will* need eval (or something like it).=0A>=A0=0A=0ASure. for =
1), things can be done most conveniently and efficiently.=0Afor 2), yeah, n=
ot sure how useful it is.=0Afor 3), maybe can let str class have a property=
 like: dy.=0Awhich can do all the dirty processing. Then we may do:=0A>>> x=
 =3D 0=0A>>> "sin($x$) =3D $sin(x)$".dy=0A'sin(0) =3D 0.0'=0A=0ANot much bu=
rden to use except for the CPU, I suppose.=0A=0A>>  You mean a translator?=
=0A> =0A> Yes. It translates your dollar-strings into something that's lega=
l=0A> Python 3.3 syntax - either calls to a function like I provided above,=
=0A> or actual embedded expressions.=0A> =0A>>  The syntax is essential for=
 compatibility.=0A>>  We must distinguish dynamic strings from common strin=
gs.=0A>>  They will live peacefully together.=0A>>  (escaping the '$' in no=
rmal strings breaks compatibility,=0A>>  and the consequence of forgetting =
to escape could be=0A>>  disastrous, so definitely not an option).=0A>> =0A=
>>  May be d" is too tiny, $"..." is easier to pick out.=0A> =0A> I don't l=
ike the use of symbols like that; can someone, glancing at=0A> your code, t=
ell whether $ is an operator, a name, or something else?=0A> The original d=
 is probably better for that.=0A>=A0=0A=0A=0AYeah, the d is probably better=
.=A0=0A=0ACheers,=0AYingjie=0A