From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on
	ip-172-31-65-14.ec2.internal
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=3.0 tests=BAYES_00,T_SCC_BODY_TEXT_LINE
	autolearn=ham autolearn_force=no version=3.4.6
Path: eternal-september.org!reader01.eternal-september.org!aioe.org!gy7opw3oYmwPg+L8gXeD2w.user.46.165.242.75.POSTED!not-for-mail
From: Stephen Leake <stephen_leake@stephe-leake.org>
Newsgroups: comp.lang.ada
Subject: Re: String view of file
Date: Mon, 02 Jan 2023 10:57:11 -0800
Organization: Aioe.org NNTP Server
Message-ID: <86h6x8pxx4.fsf@stephe-leake.org>
References: <d1d4d819-8da2-4442-8bf9-a00c7ab6b4a5n@googlegroups.com>
	<10f5dfec-32fb-4333-a9b6-2ee71c1871c0n@googlegroups.com>
Mime-Version: 1.0
Content-Type: text/plain
Injection-Info: gioia.aioe.org; logging-data="56535"; posting-host="gy7opw3oYmwPg+L8gXeD2w.user.gioia.aioe.org"; mail-complaints-to="abuse@aioe.org";
User-Agent: Gnus/5.13 (Gnus v5.13)
X-Notice: Filtered by postfilter v. 0.9.2
Cancel-Lock: sha1:yN5V1Vbzh70OEdheWu+oKciE+P8=
Xref: reader01.eternal-september.org comp.lang.ada:64766
List-Id: <comp.lang.ada>

Jesper Quorning <jesper.quorning@gmail.com> writes:

> The original post was a bit of a mess, and I am not quite new to Ada. But I lack some skills regarding software construction and software engineering.
>
> What I want goes something like this:
> - Zero copy access to a (read-only) file.

You can use gnatcoll mmap to get a memory-mapped view of a file. To get
a zero-copy string, you'll have to find the line boundaries yourself,
and use unchecked-conversion on a string pointer, or an address clause
on a local variable.

> - Get lines of file one by one as a stream of lines.

'stream' has a specific meaning in Ada; better say "collection of lines"
here.

You can define an iterator for your File type.

- Use a package instance as file representation. 

Why do you want to do this? The alternative is to define your own File
type, and provide an Open; then you can reuse that package for multiple
files. Closer to Ada standard practice for files.

- Stretch goal: Multiple implementations of File Instance (Standard, Fast, Compatible ..)

If you want to choose the implementation at compile time, you can use
separate bodies in implementation-specific directories, and choose the
directory in the project file.

If you want to chose the implementation at run time, you have to make
File a dispatching type. Then you pick the implementation via the
package name:

My_File : Files.Implementation_1.File;

> for Line of File.As_Line_Stream loop
>    Parse (Line);
> end loop;

You probably know that you can get close to this in standard Ada, at the
cost of a copy:

while not End_of_file (File) loop
   declare
      Line : get_line (file);
   begin
      Parse (line);
   end;
end loop;

An really aggressive optimizing compiler could get rid of the local
variable Line to avoid the copy, but it's not very likely.

This violates your requirement, but it could easily be that the copy is
very cheap; have you measured it in your actual application, and is
eliminating it worth the bother?

-- 
-- Stephe