RE: MFC with egcs 1.1 (Was: MFC with CygWin)

RE: MFC with egcs 1.1 (Was: MFC with CygWin)

[Sergey Okhapkin]

Thomas Nichols wrote:
> what is the status of attempts to access the MFC DLLs from egcs? If the
> name-mangling schemes are incompatible, will this remain true in
> perpetuity, or are there plans to allow MS-style name mangling as an option
> to gcc?
> 

Name mangling is not the only problem. Gcc passes "this" pointer when calling class member on top of stack, but MSVC in ecx.

-- 
Sergey Okhapkin, http://www.lexa.ru/sos
Moscow, Russia


-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

RE: MFC with egcs 1.1 (Was: MFC with CygWin)

[Thomas Nichols]

At 16:25 09/10/98 +0300, Sergey Okhapkin wrote:
>Thomas Nichols wrote:
>> what is the status of attempts to access the MFC DLLs from egcs? If the
>> name-mangling schemes are incompatible, will this remain true in
>> perpetuity, or are there plans to allow MS-style name mangling as an option
>> to gcc?
>> 
>
>Name mangling is not the only problem. Gcc passes "this" pointer when
calling class member on top of stack, but MSVC in ecx.
>
>-- 
>Sergey Okhapkin, http://www.lexa.ru/sos
>Moscow, Russia


What are the possibilities of an "MSVC compatibility" switch for gcc at
some future date? Do the compiler writers have this as a stated goal? If it
can be done, and if the licensing can be resolved, GPL'd software could be
quite as important under Windows as it has been under Un*x. (Before you
ask, sorry, but I don't have any experience of building compilers... I can
put in some time at the moment, though, if I can contribute anything useful.)

Regards,
Thomas Nichols.



-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: MFC with egcs 1.1 (Was: MFC with CygWin)

[Mumit Khan]

Thomas Nichols <thomas.nichols@iname.com> writes:
> 
> What are the possibilities of an "MSVC compatibility" switch for gcc at
> some future date? Do the compiler writers have this as a stated goal? 

Very little. Not at present. (Not speaking for the egcs team, but mostly
from looking at who's doing what).

The only feature that will be integrated in the near future is the ability
to use COM objects etc (thanks to Antonio Mendes de Oliveira Neto).

The vtable layout needs to be changed significantly, in addition to what
Sergey points out, before MFC can be handled and there may be patent
issues as well. Note that MS essentially forces other vendors to pay
licensing fees, which is obviously not going to happen in GNU software.

Regards,
Mumit
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: MFC with egcs 1.1 (Was: MFC with CygWin)

[Thomas Nichols]

At 11:28 10/10/98 -0500, Mumit Khan wrote:
>Thomas Nichols <thomas.nichols@iname.com> writes:
>> 
>> What are the possibilities of an "MSVC compatibility" switch for gcc at
>> some future date? Do the compiler writers have this as a stated goal? 
>
>Very little. Not at present. (Not speaking for the egcs team, but mostly
>from looking at who's doing what).


Thanks very much, I'll just have to get VC++ 6 then. Or maybe I'll go for
Power++ instead, even if it is stagnating, 'cos it's such a nice tool...

Thanks to all for the advice,

Regards,

Thomas.


-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Precompiled headers in egcs? (was MFC with egcs 1.1)

[Hilton Fernandes]

On 10 Oct 98 at 19:58, Thomas Nichols <thomas.nichols@iname.com> 
wrote:

> Thanks very much, I'll just have to get VC++ 6 then. Or maybe I'll go for
> Power++ instead, even if it is stagnating, 'cos it's such a nice tool...
> 
> Thanks to all for the advice,
> 
> Regards,
> 
> Thomas.
> 

Hello! 

If you're not exclusively interested in MFC, but in some OO library
that can make easier the development of GUI apps, you should take a
look at wxWindows, found in

	http://web.ukonline.co.uk/julian.smart/wxwin/

and V, that can be found in

	http://www.objectcentral.com

They're public domain software -- V is GPL --, they have active 
mailing lists.

Their codes are cleaner than MFC's: you can actually understand
them, simply by reading the code. :-)  No necessity of huge books 
like MFC Internals, just to find what was the original intention of 
the programmer when writing that piece of code. :-(

They have different intentions and code base, but they show
the same quality: portability between Win32 GUI and several Unix
GUIs, based on X Windows.  They even have some support for the Win16 
API, aka Windows 3.1X.

BTW, since we're talking about compilers, i'd like to ask whether
there will be some effort to include precompiled headers in EGCS. As
we all know, programming class libraries demand lots of headers. The
repeated recompilation of them during the development incurs in
wasted time, as they're not subjected to change.  Since Borland 3.1
(circa 1990?), compilers save the unaltered headers in a parsed
form, that shaves compilation time.

The question is: will the precompiled header feature be someday part
of EGCS?  Would it be structurally difficult, or even impossible?

Thanks for your attention.


Best regards,
--Hilton
----
Hilton Fernandes
hgfernan@usp.br
hfernandes@geocities.com
http://www.geocities.com/SiliconValley/Lakes/5657
URLs and help on C++ programming and Object-Oriented Design
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: Precompiled headers in egcs? (was MFC with egcs 1.1)

[Mumit Khan]

"Hilton Fernandes" <hgfernan@usp.br> writes:

> The question is: will the precompiled header feature be someday part
> of EGCS?  Would it be structurally difficult, or even impossible?

It'll be done as soon as someone steps forward and does the actual 
work ;-) Considering how much needs to be done before gcc is a viable
replacement for commercial x86-win32 compilers, I'd say it's a pretty
low priority item. Then again, if someone contributes the work, it'll
probably go in right away ...

Regards,
Mumit

-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: Precompiled headers in egcs? (was MFC with egcs 1.1)

[Thomas Nichols]

Hilton,

>If you're not exclusively interested in MFC, but in some OO library
>that can make easier the development of GUI apps, you should take a
>look at wxWindows, found in
>
>	http://web.ukonline.co.uk/julian.smart/wxwin/
>
>and V, that can be found in
>
>	http://www.objectcentral.com

Thanks for these, both of which I've checked out previously. wxWindows
version 2 looks quite promising, but neither of them seem to have anything
like the scope and power of Zinc at
http://www.zinc.com

This is very "clean", fully OO/event driven, has a graphical cross-platform
"resource editor" (Zinc Designer) which allows (e.g.) graphical positioning
of derived window objects and automatic code generation.

Although this is currently a commercial product, there is a "Personal
Version" of Zinc, which can be downloaded, but at present the licence (as I
read it) does not permit redistribution of the binaries you produce. Do
please write to them to urge them to put the Personal Version under the GPL
or similar, it's an ideal cross-platform tool, currently addressing Win32/X
Motif/Mac/DOS32/embedded (pSOS, VxWorks). It supports the GNU gcc under
Motif, and can be used with egcs with trivial mods (see previous postings
to this list). The Personal Version is simply the full version without the
ZafTable class (which is itself fairly amazing). The only problem with all
this is that nobody (except Peter Coffee of PC Week, who rates it very
highly) seems to have heard of it.

Rather academic from my point of view - I know Zinc, I don't know MFC, and
(in the UK) the bulk of jobs advertised are for MFC. 

Thanks,
Regards,
Tom.


-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: Precompiled headers in egcs? (was MFC with egcs 1.1)

[Don Sharp]

Hilton Fernandes wrote:
> 
> On 10 Oct 98 at 19:58, Thomas Nichols <thomas.nichols@iname.com>
> wrote:
> 
> > Thanks very much, I'll just have to get VC++ 6 then. Or maybe I'll
> go for
> > Power++ instead, even if it is stagnating, 'cos it's such a nice
> tool...
> >
> > Thanks to all for the advice,
> >
> > Regards,
> >
> > Thomas.
> >
> 
> Hello!
> 
> If you're not exclusively interested in MFC, but in some OO library
> that can make easier the development of GUI apps, you should take a
> look at wxWindows, found in
> 
>         http://web.ukonline.co.uk/julian.smart/wxwin/
> 
> and V, that can be found in
> 
>         http://www.objectcentral.com
> 
> They're public domain software -- V is GPL --, they have active
> mailing lists.
> 
> Their codes are cleaner than MFC's: you can actually understand
> them, simply by reading the code. :-)  No necessity of huge books
> like MFC Internals, just to find what was the original intention of
> the programmer when writing that piece of code. :-(
> 
> They have different intentions and code base, but they show
> the same quality: portability between Win32 GUI and several Unix
> GUIs, based on X Windows.  They even have some support for the Win16
> API, aka Windows 3.1X.
> 
> BTW, since we're talking about compilers, i'd like to ask whether
> there will be some effort to include precompiled headers in EGCS. As
> we all know, programming class libraries demand lots of headers. The
> repeated recompilation of them during the development incurs in
> wasted time, as they're not subjected to change.  Since Borland 3.1
> (circa 1990?), compilers save the unaltered headers in a parsed
> form, that shaves compilation time.
> 
> The question is: will the precompiled header feature be someday part
> of EGCS?  Would it be structurally difficult, or even impossible?
> 
> Thanks for your attention.
> 
> Best regards,
> --Hilton
> ----
> Hilton Fernandes
> hgfernan@usp.br
> hfernandes@geocities.com
> http://www.geocities.com/SiliconValley/Lakes/5657
> URLs and help on C++ programming and Object-Oriented Design
> -
> For help on using this list (especially unsubscribing), send a message
> to
> "gnu-win32-request@cygnus.com" with one line of text: "help".

My lengthy experience with pre-compiled headers, mainly with Borland
compilers, shows that unless the sequence of your include files is
always the same then the pre-compiled header database grows like topsy.
My memory is of a 54Mb precompiled header database for a ~4Mb total
source file project. Still disc space is cheap and the database lookups
may be
fast enough even when the database becomes large.

A better answer might be an automated way to remove spacing and comments
from header files and save them in a special directory. I believe that
it was Jacob Navia who reported major speed improvements as a result.

Cheers

Don Sharp

-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: Precompiled headers in egcs? (was MFC with egcs 1.1)

[Hilton Fernandes]

On 13 Oct 98 at 9:21, Don Sharp <dwsharp@iee.org> wrote:

> My lengthy experience with pre-compiled headers, mainly with Borland
> compilers, shows that unless the sequence of your include files is
> always the same then the pre-compiled header database grows like topsy.
> My memory is of a 54Mb precompiled header database for a ~4Mb total
> source file project. Still disc space is cheap and the database lookups
> may be
> fast enough even when the database becomes large.
> 

You suggest an interesting benchmark: to compile applications with 
huge include file sets, and to assess the precompiled header length 
and compilation speed with and without precompiled headers.

That comparison would have to use several commercial compilers.  If 
you want my opinion, i think Borland isn't the leader in compiler 
writing anymore, but the Redmond people. :-(  So, i'm not very 
impressed with their figures.

> A better answer might be an automated way to remove spacing and
> comments from header files and save them in a special directory. I
> believe that it was Jacob Navia who reported major speed
> improvements as a result.
> 

Navia's idea is very interesting and seem to defy a programmer's 
idea: you can use any comments and spacing that you want and still 
have the same compilation speed.

However, to automatically uncomment, remove spacing and save in a
special directory can still be considered a form of precompiled
header creation.

> Cheers
> 
> Don Sharp
> 
> -

Best regards,
--Hilton
-------------------
Hilton Fernandes, M. Sc
email: hgfernan@usp.br
Parallel Distributed Applications 
at Escola Politecnica (Polytechnic School)
University of S. Paulo - Brazil
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".