[#35036] Intentional Programming — "John" <nojgoalbyspam@...>

Hi all

17 messages 2002/03/01

[#35112] RDoc question — Michael Davis <mdavis@...>

I have a question about RDoc. I would like to reference an external

17 messages 2002/03/02

[#35162] string to array and back — Ron Jeffries <ronjeffries@...>

I am needing to convert strings to arrays of bytes and back. I see pack and

19 messages 2002/03/03

[#35364] file reading impossibly slow? — Ron Jeffries <ronjeffries@...>

So I'm doing this benchmark to work with my set program. Part of the problem is

18 messages 2002/03/07

[#35429] Interesting link on static/dynamic typing... — Robert Feldt <feldt@...>

...relevant to Ruby compared to other languages discussion:

25 messages 2002/03/08
[#35441] Re: Interesting link on static/dynamic typing... — Paul Brannan <paul@...> 2002/03/08

On Fri, Mar 08, 2002 at 05:34:43PM +0900, Robert Feldt wrote:

[#35460] Spam, ruby-talk, and me — Dave Thomas <Dave@...>

14 messages 2002/03/08

[#35537] Confusion — David Corbin <dcorbin@...>

The following is from my debugging through xmlc.rb

16 messages 2002/03/10

[#35579] RE: WIN32OLE and LDAP — "Morris, Chris" <chris.morris@...>

> The new version 0.4.2 of Win32OLE has WIN32OLE.bind method.

16 messages 2002/03/11

[#35652] Method type 'abstract' — Peter Hickman <peter@...>

The one thing I miss in Ruby is the abstract class method to go along

15 messages 2002/03/12

[#35653] Some potential RCRs — "Bob Alexander" <bobalex@...>

Here are a few thing I am considering submitting as RCRs. I'm looking =

50 messages 2002/03/12
[#35672] Re: Some potential RCRs — matz@... (Yukihiro Matsumoto) 2002/03/12

Hi,

[#35683] Re: Some potential RCRs — Massimiliano Mirra <list@...> 2002/03/12

On Wed, Mar 13, 2002 at 03:58:01AM +0900, Yukihiro Matsumoto wrote:

[#35697] Re: Some potential RCRs — David Alan Black <dblack@...> 2002/03/13

Hello --

[#35694] rpkg 0.3 — Massimiliano Mirra <list@...>

14 messages 2002/03/13
[#35699] RE: [ANN] rpkg 0.3 — <james@...> 2002/03/13

>

[#35787] testunit - setup -> set_up ? — "Morris, Chris" <chris.morris@...>

I'm just starting to use testunit instead of rubyunit ... I noticed with an

21 messages 2002/03/13
[#35793] RE: testunit - setup -> set_up ? — "Nathaniel Talbott" <nathaniel@...> 2002/03/13

Morris, Chris [mailto:chris.morris@snelling.com] wrote:

[#35796] Re: testunit - setup -> set_up ? — Dave Thomas <Dave@...> 2002/03/13

"Nathaniel Talbott" <nathaniel@talbott.ws> writes:

[#35797] RE: testunit - setup -> set_up ? — "Nathaniel Talbott" <nathaniel@...> 2002/03/13

dave@thomases.com [mailto:dave@thomases.com] wrote:

[#35898] camelCase and underscore_style — "Morris, Chris" <chris.morris@...>

First, a question. If underscore_style is the Ruby norm for methods and the

20 messages 2002/03/15
[#35924] Re: camelCase and underscore_style — "Guy N. Hurst" <gnhurst@...> 2002/03/15

Phil Tomson wrote:

[#35930] RE: camelCase and underscore_style — "Nathaniel Talbott" <nathaniel@...> 2002/03/16

Guy N. Hurst [mailto:gnhurst@hurstlinks.com] wrote:

[#35989] ANN: Locana GUI and GUI Builder version 0.81 — Michael Davis <mdavis@...>

I am pleased to announce release 0.81 of Locana. Locana is a GUI

16 messages 2002/03/16

[#35992] XPath — Michael Schuerig <schuerig@...>

27 messages 2002/03/16

[#36034] Mini Rant: Indenting — Thomas Hurst <tom.hurst@...>

Why is it that I see *so* much code like:

14 messages 2002/03/17

[#36049] web templating for static sites? — Massimiliano Mirra <list@...>

I'm using the Template Toolkit for generating static web sites and I

42 messages 2002/03/17
[#36426] web standars (was: web templating for static sites?) — Tobias Reif <tobiasreif@...> 2002/03/20

Albert Wagner wrote:

[#36052] Xml Serialization for Ruby — "Chris Morris" <chrismo@...>

=Xml Serialization for Ruby

20 messages 2002/03/17
[#36059] Re: [ANN] Xml Serialization for Ruby — Massimiliano Mirra <list@...> 2002/03/17

On Mon, Mar 18, 2002 at 05:20:56AM +0900, Chris Morris wrote:

[#36067] eval/Module question — David Corbin <dcorbin@...>

If I have a String src that is similar to the following:

13 messages 2002/03/18

[#36157] Development of Windows version of Ruby — ptkwt@...1.aracnet.com (Phil Tomson)

Now that we've dumped the cygwin requirement for the Windows version of

63 messages 2002/03/18
[#36330] Re: Development of Windows version of Ruby — Ron Jeffries <ronjeffries@...> 2002/03/19

On Tue, 19 Mar 2002 14:05:27 GMT, "Albert L. Wagner" <alwagner@uark.edu> wrote:

[#36431] Re: Development of Windows version of Ruby — Dennis Newbold <dennisn@...> 2002/03/20

[#36458] Windows version of Ruby (proposals) — ptkwt@... (Phil Tomson) 2002/03/21

Dennis Newbold <dennisn@pe.net> wrote in message news:<Pine.GSO.3.96.1020320113603.22242B-100000@shell2>...

[#36482] RE: Windows version of Ruby (proposals) — "Christian Boos" <cboos@...> 2002/03/21

Some thoughts on the 2 first Windows issues, plus a 4th one...

[#36496] Re: Windows version of Ruby (proposals) — Dave Thomas <Dave@...> 2002/03/21

"Christian Boos" <cboos@bct-technology.com> writes:

[#36510] Re: Windows version of Ruby (proposals) — nobu.nokada@... 2002/03/21

Hi,

[#36514] Re: Windows version of Ruby (proposals) — Dave Thomas <Dave@...> 2002/03/21

nobu.nokada@softhome.net writes:

[#36518] Re: Windows version of Ruby (proposals) — nobu.nokada@... 2002/03/21

Hi,

[#36211] dots in Dir.entries — matz@... (Yukihiro Matsumoto)

Hi,

22 messages 2002/03/19

[#36231] style choice — Ron Jeffries <ronjeffries@...>

A style question for the community ... which of the following do you prefer, and

18 messages 2002/03/19

[#36345] ANN: REXML 2.0 — Sean Russell <ser@...>

I have a feeling there will only be three major revisions of REXML. Version

19 messages 2002/03/20

[#36610] Re: Windows version of Ruby (proposals) — Ron Jeffries <ronjeffries@...>

On Thu, 21 Mar 2002 14:11:55 GMT, Dave Thomas <Dave@PragmaticProgrammer.com> wrote:

16 messages 2002/03/22

[#36645] Ruby for Mac OS 10.1 — Jim Freeze <jim@...>

Hi:

28 messages 2002/03/23

[#36768] Re: Difference between 'do' and 'begin' — Clemens Hintze <c.hintze@...>

In <slrna9ulvi.f2h.mwg@fluffy.isd.dp.ua> Wladimir Mutel <mwg@fluffy.isd.dp.ua> writes:

23 messages 2002/03/26
[#36783] RE: Difference between 'do' and 'begin' — <james@...> 2002/03/26

[#36792] Re: Difference between 'do' and 'begin' — Kent Dahl <kentda@...> 2002/03/26

james@rubyxml.com wrote:

[#36808] Error calling Tk in a loop — <james@...>

I'm trying to write some code that pops up a Tk window when for certain

15 messages 2002/03/26

[#36841] RE: Windows version of Ruby (proposals) — "Andres Hidalgo" <sol123@...>

I believe that Ruby has a place in windows (Office), I happened to have

14 messages 2002/03/27

[#36863] Hash.new(Hash.new) doesn't use Hash.new as default value — "Jonas Delfs" <jonas@...>

Hi -

18 messages 2002/03/27

[#37080] Why isn't Math object-oriented? — Bil Kleb <W.L.Kleb@...>

So I'm reading along in the Pixaxe book (yet again), and I am told

15 messages 2002/03/30

[#37121] String#begins?(s) — timsuth@... (Tim Sutherland)

class String

24 messages 2002/03/31

Re: Windows version of Ruby (proposals)

From: Dave Thomas <Dave@...>
Date: 2002-03-21 14:04:55 UTC
List: ruby-talk #36496
"Christian Boos" <cboos@bct-technology.com> writes:

> > Anyway, Apparently popen works on 1.6.6 so that leaves only fork not
> > working (is there anything else?).
> 
> Ok, popen works, but not Open3::popen3.

Does popen work with "w+" in your version of Ruby under Windows?

I'd add a number of other issues.

- Threading doesn't work correctly
- $? doesn't work in all circumstances
- File mode stuff is strange
- the 'cmd /c' prefixing of stuff run with system is inconsistent

and so on.

However, I'm not sure that I'd want to see (say) 'fork' emulated under
Windows.

I have a vision for the libraries in 2.0 which I'm sure will never see
the light of day.

I'm thinking that there's stuff that is purely Ruby, and there's stuff
that is environment-sensitive. For example, class Hash is class Hash
no matter where it's running. It should be universally available in
all Ruby implementations.

However, class File is different.

    File.open(name, modestring) do |f|
      line = f.read
    end

is universal - it should work on every Ruby platform. So are many
other File methods. However

    File.open(name, modenum, permnum)

is _not_ portable, and so should not be available universally.
Similarly, File.join is portable (if broken under Windows...),
File.lstat is not. File.delete is OK, File.mtime is not.

And then, under Windows, there are a whole lot of missing file
methods, methods that let us access the underlying Win32 calls for
events, ACLs etc.

So, here's my thinking. The basic classes you get with Ruby (Array,
Hash, Dir, File, Process etc) should *only* contain methods that are
guaranteed to be portable across all Ruby's platforms. If you code
using just these, your program will run unmodified and consistently
anywhere that Ruby runs.

Then, we provide a set of environment support modules for those
classes that currently contain non-portable methods. We could have
(for example)

    module SysdepFilePosix
      ...
    end

    module SysdepFileWin32
      ...
    end

Mix one of these into class File and you'll get all the system
dependent calls. You'll also be explicitly marking your code as
non-portable.

Then, to make life easier, you have a global set of files, say
sysdep_posix, sysdep_win32, sysdep_macosx etc, whose whole purpose is
to mix in the system dependent modules into _all_ the classes that
need them.

That way, you'd have control. If you just wanted to add Window's
specific stuff to File, you could do

   class File
     include SysdepFileWin32
   end

If instead you were writing a program which needed lots of Windows
specific features, you'd do

   require 'sysdep_win32'

and _all_ the appropriate classes would be modified (it might even add
in whole new classes, such as Win32OLE).


I think I'd then take this a step further and add a new method to
Kernel called 'sysdep'

   sysdep :win32
      would be the same as 'require "sysdep_win32"'

   sysdep :win32, File, Dir
      would be the same as

        class File
           include SysdepFileWin32
        end

        class Dir
           include SysdepDirWin32
        end

'sysdep' would check that the platform you're running on is compatible
with the extensions you ask for, raising an exception if not.

Using this scheme, we make portability an explicit feature of the
language, rather then trying to impose a half-hearted Unix layer on
everyone.


Cheers


Dave

In This Thread