From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00 autolearn=ham autolearn_force=no version=3.4.4 X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 103376,2d76c4749a82ef0b X-Google-Attributes: gid103376,public X-Google-ArrivalTime: 2003-10-27 11:37:22 PST Path: archiver1.google.com!news2.google.com!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!logbridge.uoregon.edu!hammer.uoregon.edu!skates!not-for-mail From: Stephen Leake Newsgroups: comp.lang.ada Subject: Re: Strings Fixed, Bounded, and Counted. Performance Inquiry. Date: 27 Oct 2003 14:31:23 -0500 Organization: NASA Goddard Space Flight Center (skates.gsfc.nasa.gov) Message-ID: References: NNTP-Posting-Host: shevek.gsfc.nasa.gov Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: skates.gsfc.nasa.gov 1067283304 15872 128.183.235.101 (27 Oct 2003 19:35:04 GMT) X-Complaints-To: usenet@news.gsfc.nasa.gov NNTP-Posting-Date: 27 Oct 2003 19:35:04 GMT User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 Xref: archiver1.google.com comp.lang.ada:1747 Date: 2003-10-27T19:35:04+00:00 List-Id: Freejack writes: > On Mon, 27 Oct 2003 12:18:48 -0500, Stephen Leake wrote: > > > >> A bit more flexible than Fixed strings, > > > > How? > > It was my understanding(I could be wrong here) that a Fixed string is > unmodifiable without access to functions provided in the package. A > Counted string is modifiable outside the package provided functions and > procedures. No? If you mean an Ada Counted string, you get to define it. If that's what you want, so be it. I don't see why using "functions provided in the package" is a problem; that's what they are for! > >> a heck of a lot faster than Null terminated strings. > > > > Depends on what you are doing. Copying the strings character by > > character is the same speed for either construct. Block copy is faster > > with a count on most machines. But most algorithms for processing > > strings operate character by character, so it doesn't really matter. > > > > Safety is another issue; with a count (or bounds) it is possible to > > check array indices before fetching or storing. > > So essentially, with a Fixed string, I can have the compiler generate > machine code to grab a block of length X, and shove it into an empty > block of length Y if Y > X? With any Ada string, Fixed or not. Yes. > All on one I/O procedure? (Thinking of the Linux "sendfile()" and > "iovec()" functions here.) I really want to avoid character by > character transfer when it's not necessary. Let the compiler worry about that. > Also, if I can read the first element, and see that the string has > twenty characters, I can have the program inline a procedure x/20 > times, rather than using a Counter and loop structure. That's what 'Length is for. Or the Length function in the Ada.Strings.* packages. They are all very fast functions; you should not worry about how fast they are. > >> I'm gonna write my own experimental package if nobody else has > > > > Be sure you define _precisely_ what you want the package to do, and see > > if Ada already has it, before you spend time implementing it. > > I am definitely being specific in my package spec. You're right. Aside > from a couple quirks of my own design, bounded strings do the right > thing. As do Fixed Strings. > > I'll keep playing with 'em and tell you what I come up with. Heh. Ok, sounds good. -- -- Stephe