tom@ssd.csd.harris.com (Tom Horsley) (08/22/90)
This is really two questions, first the most general one:
Is there a really complete description of the new iostream interface in a
easily available book or document of some kind out there?
(I have seen a reference to an official AT&T document, is it reasonably
priced, and can I order a copy from some 800 number or something (or even
FTP it)?).
Second, the specific one that lead to my asking the first:
How do you determine the number of bytes actually read by the read()
function in an iostream?
I have just started using TurboC++ and looking through the <iostream.h>
header I found a gcount() function with a comment that says it returns the
number of bytes read by the last unformatted input operation. Since read()
is an unformatted input operation, I would have expected this to be the
function to use, but when I do several consecutive read() calls, the value
returned by gcount() is the cumulative total of all the bytes read in all
the calls.
Is gcount() the correct official function to use to determine the size
of the last read?
Is the TurboC++ iostream implementation broken?
I have a couple of C++ books, but none of them go into any real detail on
iostreams (although they all mention them in passing).
--
======================================================================
domain: tahorsley@csd.harris.com USMail: Tom Horsley
uucp: ...!uunet!hcx1!tahorsley 511 Kingbird Circle
Delray Beach, FL 33444
+==== Censorship is the only form of Obscenity ======================+
| (Wait, I forgot government tobacco subsidies...) |
+====================================================================+daniel@terra.ucsc.edu (Daniel Edelson) (08/24/90)
In article <TOM.90Aug22070026@hcx2.ssd.csd.harris.com> tom@ssd.csd.harris.com (Tom Horsley) writes: > Is there a really complete description of the new iostream interface in a > easily available book or document of some kind out there? > > (I have seen a reference to an official AT&T document, is it reasonably > priced, and can I order a copy from some 800 number or something (or even > FTP it)?). I believe the document you are interested in is: UNIX System V AT&T C++ Language System Release 2.1 Library Manual Select Code 307-158 This document contains the IOSTREAM(3C++) manual pages. Daniel Edelson daniel@cis.ucsc.edu and ...!sun!practic!peren!daniel (Perennial Software, Inc.)
sarima@tdatirv.UUCP (Stanley Friesen) (08/24/90)
And he's not the only one. I have been looking, without much luck, for some time. Sigh, Zortech, in their infinite wisdom, decided NOT to use the iostream library in version 2.0 of their compiler! So even with a supposedly 2.0 compatible C++ compiler I am stuck with the old 1.2 streams. (P.S. does the new update from Zortech contain the iostream library?)
bright@Data-IO.COM (Walter Bright) (08/25/90)
In article <127@tdatirv.UUCP> sarima@tdatirv.UUCP (Stanley Friesen) writes:
<Sigh, Zortech, in their infinite wisdom, decided NOT to
<use the iostream library in version 2.0 of their compiler! So even with
<a supposedly 2.0 compatible C++ compiler I am stuck with the old 1.2 streams.
Let me explain our rationale behind this. First, iostreams is not part of
the C++ standard as defined by the ARM. The ARM does not mention iostreams.
Second, a large part of the functionality of iostreams is contained in
the header files (one tendency of C++!). A large part of the functionality
is also not available as a specification. This means that in order to
clone iostreams, we'd have to largely plagarize AT&T's code, which is
copyrighted. Trying to figure out a non-infringing same-only-different
is an extremely difficult task, and we do not have the resources to
do a 'clean room' approach like Phoenix did with the IBM BIOS. We do
not wish to steal AT&T's proprietary code.
Much to AT&T's credit, they have taken a very friendly attitude towards
this. Jerry Schwarz, author of the original iostreams, has undertaken
writing a specification for a cleaned up and simplified iostreams for
eventual inclusion into the ANSI C++ standard. This is with AT&T's full
support. (I wish more companies behaved like this instead of engaging
in lawyer enrichment!)
Zortech is committed to fully supporting the evolving ANSI C++ standard
and will implement an iostreams to Jerry's new spec.
<(P.S. does the new update from Zortech contain the iostream library?)
Nodl@g.g.oswego.edu (Doug Lea) (08/26/90)
Walter Bright says: > [...] > First, iostreams is not part of > the C++ standard as defined by the ARM. The ARM does not mention iostreams. > Second, a large part of the functionality of iostreams is contained in > the header files (one tendency of C++!). A large part of the functionality > is also not available as a specification. This means that in order to > clone iostreams, we'd have to largely plagarize AT&T's code, which is > copyrighted. > [...] > We do not wish to steal AT&T's proprietary code. > [...] > Zortech is committed to fully supporting the evolving ANSI C++ standard > and will implement an iostreams to Jerry's new spec. (or rather, the ANSI libraries committee's new spec). Otherwise, ditto for the GNU C++ library (libg++). One of the most interesting questions about all this is the impact of exception support on the design and implementation of stream classes. -Doug -- Doug Lea, Computer Science Dept., SUNY Oswego, Oswego, NY, 13126 (315)341-2688 email: dl@g.oswego.edu or dl@cat.syr.edu UUCP :...cornell!devvax!oswego!dl or ...rutgers!sunybcs!oswego!dl
sarima@tdatirv.UUCP (Stanley Friesen) (08/27/90)
In article <2667@dataio.Data-IO.COM> bright@Data-IO.COM (Walter Bright) writes: >Second, a large part of the functionality of iostreams is contained in >the header files (one tendency of C++!). A large part of the functionality >is also not available as a specification. This means that in order to >clone iostreams, we'd have to largely plagarize AT&T's code, which is >copyrighted. Trying to figure out a non-infringing same-only-different >is an extremely difficult task, and we do not have the resources to >do a 'clean room' approach like Phoenix did with the IBM BIOS. We do >not wish to steal AT&T's proprietary code. Thanks for the response. This, unfortunately, still leaves me in the unpleasant position of witing code that is already essentially out of date; it will simply not be portable to other version 2.0 C++ compilers. Considering that I am working (in my limited spare time) on a project I would like to sell, this is a serious limitation. So, to all of you out there in netland - is there any way I can get a copy of those blasted header files for iostreams!!!! Then I could at least write my code in a portable manner, by implementing the rest of it myself. (This btw is why I have been looking for a good book on iostreams, but even the best I have found yet is not good enough to write a matching implementation - it simply does not cover some of the advanced features like (oh blast I forget what they are called - you know, those functionoids you insert into a stream to change its internal state during I/O)). >Zortech is committed to fully supporting the evolving ANSI C++ standard >and will implement an iostreams to Jerry's new spec. In general I have been quite pleased with you support of this. ><(P.S. does the new update from Zortech contain the iostream library?) > >No Sigh, oh well, I shall just have to wait for 2.2 before I spend money on an upgrade. ---------------------------- uunet!tdatirv!sarima (aka Stanley Friesen)