From 6689203760242465871 X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f78e5,b83a4267f4ac775 X-Google-Attributes: gidf78e5,public From: Valentin Bonnard Subject: Re: string/cow/algorithm Date: 1998/08/11 Message-ID: <35D0B753.1C9D@pratique.fr>#1/1 X-Deja-AN: 380207557 X-NNTP-Posting-Host: cyber93-9.paris.imaginet.fr Content-Transfer-Encoding: 7bit Approved: stephen.clamage@sun.com (comp.std.c++) References: <6qfr41$oue$1@pigpen.csrlink.net> <6qg3ld$r7o$1@shell7.ba.best.com> <6qgs0r$92e$1@pigpen.csrlink.net> <35ccc2b4.832003@news3.ibm.net> <6qn19u$sb0$1@pigpen.csrlink.net> <6qnj99$h2l$1@nnrp1.dejanews.com> <6qo3fq$dum$1@pigpen.csrlink.net> X-UID: 0000000001 X-Status: $$$T Content-Type: text/plain; charset=us-ascii Organization: Ecole Normale Superieure, Paris, France Mime-Version: 1.0 Reply-To: bonnardv@pratique.fr Newsgroups: comp.std.c++ Originator: clamage@taumet John Potter wrote: > > The original question was the conformance of a string class which > sorted the original string when a copy was sorted. The answer was > that the FDIS is different from the CD2 and the implementation is not > conformant. A following question asked about making a cow string > class conformant. > > On writing a standard conformant string class. > > AllanW@my-dejanews.com wrote: > > : I think that using a proxy class should handle this fairly well Not really > Sorry. Unless this has also changed from the CD2, it is required that > string::operator[] return a T& and that string::iterator::operator* > return a T&. Proxies are expressly forbidden by these requirements. You are perfectly right. > However, I am interested in any standard > conformant techniques that may help Mark do a better job than the > little relayed information which I posted. Or maybe there are no > better techniques. The user can get a reference or pointer to the one charT in a basic_string. You have *no way* to remplace these with UDT, change their behaviour or be told when they go out of scope. So once you have released them, you are unsharable as long as they are valid. If iterator is charT*, then begin() and end() put you in the unsharable state. If iterator is a UDT, then you have to mark the string as unsharable when the iterator is dereferenced (or before). Few programmers call begin() or end() just for fun; most will dereference the iterator at some point, generally just after. So what's the point ? The point is that there will be an unsharable state in a cow basic_string. The question is how many strings are in this state, and is there still a significant gain, or a gain at all (it's application dependant, of course). Using const std::string& as arguments is still a good idea. The compiler can optimise std::string return values. There are still cases where you copy a string without touch-ing it. -- Valentin Bonnard mailto:bonnardv@pratique.fr info about C++/a propos du C++: http://pages.pratique.fr/~bonnardv/ [ comp.std.c++ is moderated. To submit articles, try just posting with ] [ your news-reader. If that fails, use mailto:std-c++@ncar.ucar.edu ] [ --- Please see the FAQ before posting. --- ] [ FAQ: http://reality.sgi.com/austern_mti/std-c++/faq.html ]