Menu
▾
▴
opencxx-users
You can subscribe to this list here.
| 2002 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
(9) |
Oct
(16) |
Nov
(14) |
Dec
(24) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2003 |
Jan
(14) |
Feb
(57) |
Mar
(72) |
Apr
(37) |
May
(21) |
Jun
(12) |
Jul
(16) |
Aug
(33) |
Sep
(24) |
Oct
|
Nov
(10) |
Dec
(8) |
| 2004 |
Jan
(6) |
Feb
(14) |
Mar
(47) |
Apr
(41) |
May
(16) |
Jun
(31) |
Jul
(78) |
Aug
(62) |
Sep
(99) |
Oct
(43) |
Nov
(35) |
Dec
(9) |
| 2005 |
Jan
(19) |
Feb
(22) |
Mar
(7) |
Apr
|
May
(5) |
Jun
(4) |
Jul
(2) |
Aug
(9) |
Sep
(15) |
Oct
(23) |
Nov
(2) |
Dec
(20) |
| 2006 |
Jan
|
Feb
(2) |
Mar
(7) |
Apr
|
May
|
Jun
(8) |
Jul
(15) |
Aug
(1) |
Sep
(4) |
Oct
|
Nov
(9) |
Dec
|
| 2007 |
Jan
|
Feb
|
Mar
|
Apr
(9) |
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2008 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(2) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(5) |
Nov
|
Dec
|
| 2009 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(2) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2010 |
Jan
|
Feb
(4) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
(2) |
Oct
|
Nov
|
Dec
|
| 2011 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(11) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: stan <st...@uc...> - 2003-02-24 23:06:56
|
is there a way to translate global-scoped declarations such as: enums, typedefs, global variables, etc using OpenC++? for global functions there is a SetDefaultMetaclassForFunctions() is there a similar system for global declarations? what about function prototypes in global scope? thank you, stan |
|
From: Alexandre T. <kt...@fr...> - 2003-02-24 17:15:17
|
Grzegorz, > You have been added to the project. Great :) > >> I'll soon be ready to checkin Michael's Gcc 3 fix to Occ. I need some >> time to read the dox on Sf about Ssh and Cvs, I haven't used that =20 >> stuff >> for a while. > > Let me know if I can be of any help. I tried to login with cvs and execute the rtag command but it fails =20 when I enter my password. I assume this is the same password I used =20 when I created my account on Sf, right ? Moreover, when executing the command "cvs rtag -b <sandbox>", a module =20 name must be provided. What should it be ? There's nothing about it in =20 your Html doc... Thanks. Alex ------------------------------------------------------------------------=20 - Alexandre Tolmos E-mail:=A0...@fr... ICQ: 92964905 ------------------------------------------------------------------------=20 - "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn." ------------------------------------------------------------------------=20 - |
|
From: James M. D. <mdu...@ya...> - 2003-02-24 12:22:05
|
--- Grzegorz Jakacki <ja...@he...> wrote: > > OpenC++ 2.6 has been released. > Congrats Grzegorz!!! I am very sorry that I have been so little help, but my time plan has been very tight. I would like to help port this to gcc 3.2, and beyond that, my plan to implement an introspector interface as a walker into OpenC++ has become more concrete. I feel that it can easily be implemented as a customized walker that walks the tree looking for ASTS to convert. mike ===== James Michael DuPont http://introspector.sourceforge.net/ __________________________________________________ Do you Yahoo!? Yahoo! Tax Center - forms, calculators, tips, more http://taxes.yahoo.com/ |
|
From: Grzegorz J. <ja...@he...> - 2003-02-24 03:25:28
|
OpenC++ 2.6 has been released. Best regards Grzegorz ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Grzegorz J. <ja...@he...> - 2003-02-24 01:15:31
|
On Sun, 23 Feb 2003, Alexandre Tolmos wrote: [snip] > > I attached the log files the testsuite produced. Testsuite cannot find shared libgc. Let's get back to this once we are done with gcc-3.x . Best regards Grzegorz > > Alex > > ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Grzegorz J. <ja...@he...> - 2003-02-24 01:13:50
|
On Fri, 21 Feb 2003, Alexandre Tolmos wrote: > Hi Grzegorz, > > > Le jeudi, 20 f=E9v 2003, =E0 02:42 Europe/Paris, Grzegorz Jakacki a =E9= crit : > > > On Wed, 19 Feb 2003, Alexandre Tolmos wrote: > > > >> Grzegorz, > >> > >> I merged Bruno's work with the current Cvs tree. > > > > Great. (Isn't his name Michael Buro ?) > > > >> The namespace problem > >> is gone but I'm facing now another problem relating to dynamic libs: > > > > Could you check in what you have before you move on to fixing > > things with Darwin port? It would be better to have two checkins, > > each covering a smaller change, than a big one with everything lumped > > together. > > > > I am attaching checkin instructions at the end of this e-mail. Should > > you encouter any problems, let me know. Also send me you SF user id. > > I created an account on Sf; my user name is kthulhu. You have been added to the project. > I'll soon be ready to checkin Michael's Gcc 3 fix to Occ. I need some > time to read the dox on Sf about Ssh and Cvs, I haven't used that stuff > for a while. Let me know if I can be of any help. Best regards Grzegorz ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Grzegorz J. <ja...@he...> - 2003-02-24 00:26:55
|
On Fri, 21 Feb 2003, James Michael DuPont wrote: > > --- Grzegorz Jakacki <ja...@he...> wrote: > > On Thu, 20 Feb 2003, [iso-8859-1] James Michael DuPont wrote: > > > > > Gregorz, > > > I am very sorry that I have not found time to do much of the > > promised > > > help. I am in between jobs, and have many things to do, very little > > > time. > > > I can commit my debian changes if you like, and also do the network > > > heavy cvs stuff, but I cannot spend much time on > > bugfixing/tracking. > > > > No problem. We can release .deb once you are ready with it. > > I will commit the debian stuff to the cvs, 2_6 branch, it just goes > into a subdir. > > > > > > > > > > > > --- Grzegorz Jakacki <ja...@he...> schrieb: > > > > > Hi, > > > > > > > > In anticipation of checkin from Alexandre Tolmos I created > > rel_2_6 > > > > branch > > > > out of current HEAD. This is what is going to end up as 2.6 > > release > > > > tarball. > > > > > > > > Also please remember, that today is the last chance to check if > > > > > > > > > > > > > > opencxx.sourceforge.net/release-2.6-testing/opencxx-2.6-exp1.tar.gz > > > > > > I have checked the current cvs 2.6 > > > cvs co -r rel_2_6 opencxx > > > > > > here are the compiler errors : > > [...] > > > > What compiler, what platform? > Debian GNU/Linux i386 unstable , gcc 3.2 OpenC++ 2.6 does not compile on gcc-3.x and is not intended to. Alexandre Tolmos is working on that now, it will go into the next release. Thanks Grzegorz > mike > > ===== > James Michael DuPont > http://introspector.sourceforge.net/ > > __________________________________________________ > Do you Yahoo!? > Yahoo! Tax Center - forms, calculators, tips, more > http://taxes.yahoo.com/ > > ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Alexandre T. <kt...@fr...> - 2003-02-23 13:54:28
|
Le vendredi, 21 f=E9v 2003, =E0 01:20 Europe/Paris, Grzegorz Jakacki a=20=
=E9crit :
> On Thu, 20 Feb 2003, Alexandre Tolmos wrote:
>
>>
>> Le jeudi, 20 f=E9v 2003, =E0 02:42 Europe/Paris, Grzegorz Jakacki a =
=E9crit=20
>> :
>>
>>> On Wed, 19 Feb 2003, Alexandre Tolmos wrote:
>>>
>>>> Grzegorz,
>>>>
>>>> I merged Bruno's work with the current Cvs tree.
>>>
>>> Great. (Isn't his name Michael Buro ?)
>>
>> Yes, sorry :)
>>
>>>
>>>> The namespace problem
>>>> is gone but I'm facing now another problem relating to dynamic =
libs:
>>>
>>> Could you check in what you have before you move on to fixing
>>> things with Darwin port? It would be better to have two checkins,
>>> each covering a smaller change, than a big one with everything =
lumped
>>> together.
>>>
>>> I am attaching checkin instructions at the end of this e-mail. =
Should
>>> you encouter any problems, let me know. Also send me you SF user id.
>>>
>>>> "metaclass.h" includes some "ltdl.h" header file which does not=20
>>>> exist
>>>> on Mosx (well, Darwin actually).
>>>
>>> This is file from libtool and it should be present in
>>> 'opencxx/libltdl/'.
>>> Is it there?
>>>
>>>> Is there a standard way to handle Dlls
>>>> across multiple Unix variants (Posix?).
>>>
>>> Yes. Libtool hides all the plarform specific stuff from you. I found
>>> "darwin" in several places in libtool script, so it looks like this
>>> system
>>> is handled.
>>>
>>>> Moreover, Dlls have a ".dylib"
>>>> extension on Darwin, not ".so" like on FreeBSD (although Darwin is=20=
>>>> Bsd
>>>> based).
>>>
>>> I also fount 'dylib' in libtool sources, so this should be covered
>>> too. If
>>> you want to give an idea of how libtool works have a look at info=20
>>> docs
>>> ('info libtool') and Autobook (aka. Goat Book) at
>>> http://sources.redhat.com/autobook/ .
>>>
>>>> Any help is welcome :)
>>>
>>> Hope this helps.
>>
>> Ok. Everything compiles now but all the tests in "testsuite" fail.
>
> Please send me all '*.log' files you can find in ./testsuite and =
below.
>
>>
>> Moreover, I have still a problem with "configure" stuff: I have to =
run
>> "configure" and "make" in the Gc directory first, then run =
"configure"
>> and "make" in the main directory. Maybe it's because I replaced the=20=
>> old
>> Gc with the latest version (6.1) but kept the original
>> "Makefile.gc.in".
>
> Most likely.
>
>> I don't know how to deal with those configuration
>> files, I need some help there :)
>
> No problem. I will have a look once you commit your changes.
>
>> Anyway, I'll try to checkin the changes made to Occ itself (the Gcc 3
>> fix).
>
> Let's focus on this first. Send me your sourceforge id, so I can add=20=
> you to the project.
>
My id is kthulhu.
I attached the log files the testsuite produced.
Alex
|
|
From: Grzegorz J. <ja...@he...> - 2003-02-22 16:30:00
|
On Thu, 20 Feb 2003, Alexandre Tolmos wrote:
>
> Le jeudi, 20 f=E9v 2003, =E0 02:42 Europe/Paris, Grzegorz Jakacki a =E9=
crit :
>
> > On Wed, 19 Feb 2003, Alexandre Tolmos wrote:
> >
> >> Grzegorz,
> >>
> >> I merged Bruno's work with the current Cvs tree.
> >
> > Great. (Isn't his name Michael Buro ?)
>
> Yes, sorry :)
>
> >
> >> The namespace problem
> >> is gone but I'm facing now another problem relating to dynamic libs:
> >
> > Could you check in what you have before you move on to fixing
> > things with Darwin port? It would be better to have two checkins,
> > each covering a smaller change, than a big one with everything lumped
> > together.
> >
> > I am attaching checkin instructions at the end of this e-mail. Should
> > you encouter any problems, let me know. Also send me you SF user id.
> >
> >> "metaclass.h" includes some "ltdl.h" header file which does not exis=
t
> >> on Mosx (well, Darwin actually).
> >
> > This is file from libtool and it should be present in
> > 'opencxx/libltdl/'.
> > Is it there?
> >
> >> Is there a standard way to handle Dlls
> >> across multiple Unix variants (Posix?).
> >
> > Yes. Libtool hides all the plarform specific stuff from you. I found
> > "darwin" in several places in libtool script, so it looks like this
> > system
> > is handled.
> >
> >> Moreover, Dlls have a ".dylib"
> >> extension on Darwin, not ".so" like on FreeBSD (although Darwin is B=
sd
> >> based).
> >
> > I also fount 'dylib' in libtool sources, so this should be covered
> > too. If
> > you want to give an idea of how libtool works have a look at info doc=
s
> > ('info libtool') and Autobook (aka. Goat Book) at
> > http://sources.redhat.com/autobook/ .
> >
> >> Any help is welcome :)
> >
> > Hope this helps.
>
> Ok. Everything compiles now but all the tests in "testsuite" fail.
Please send me all '*.log' files you can find in ./testsuite and below.
>
> Moreover, I have still a problem with "configure" stuff: I have to run
> "configure" and "make" in the Gc directory first, then run "configure"
> and "make" in the main directory. Maybe it's because I replaced the old
> Gc with the latest version (6.1) but kept the original
> "Makefile.gc.in".
Most likely.
> I don't know how to deal with those configuration
> files, I need some help there :)
No problem. I will have a look once you commit your changes.
> Anyway, I'll try to checkin the changes made to Occ itself (the Gcc 3
> fix).
Let's focus on this first. Send me your sourceforge id, so I can add you =
to the project.
Best regards
Grzegorz
>
> Alex
>
> -----------------------------------------------------------------------=
-
> -
> Alexandre Tolmos
> E-mail:=A0...@fr...
> ICQ: 92964905
> -----------------------------------------------------------------------=
-
> -
> "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn."
> -----------------------------------------------------------------------=
-
> -
>
>
>
>
##################################################################
# Grzegorz Jakacki Huada Electronic Design #
# Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang #
# tel. +86-10-64365577 x2074 Beijing 100015, China #
# Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. #
##################################################################
|
|
From: Alexandre T. <kt...@fr...> - 2003-02-21 17:32:13
|
Hi Grzegorz, Le jeudi, 20 f=E9v 2003, =E0 02:42 Europe/Paris, Grzegorz Jakacki a =E9cr= it : > On Wed, 19 Feb 2003, Alexandre Tolmos wrote: > >> Grzegorz, >> >> I merged Bruno's work with the current Cvs tree. > > Great. (Isn't his name Michael Buro ?) > >> The namespace problem >> is gone but I'm facing now another problem relating to dynamic libs: > > Could you check in what you have before you move on to fixing > things with Darwin port? It would be better to have two checkins, > each covering a smaller change, than a big one with everything lumped > together. > > I am attaching checkin instructions at the end of this e-mail. Should > you encouter any problems, let me know. Also send me you SF user id. I created an account on Sf; my user name is kthulhu. I'll soon be ready to checkin Michael's Gcc 3 fix to Occ. I need some =20 time to read the dox on Sf about Ssh and Cvs, I haven't used that stuff =20 for a while. Regards, Alex ------------------------------------------------------------------------=20 - Alexandre Tolmos E-mail:=A0...@fr... ICQ: 92964905 ------------------------------------------------------------------------=20 - "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn." ------------------------------------------------------------------------=20 - |
|
From: James M. D. <mdu...@ya...> - 2003-02-21 10:24:18
|
--- Grzegorz Jakacki <ja...@he...> wrote: > On Thu, 20 Feb 2003, [iso-8859-1] James Michael DuPont wrote: > > > Gregorz, > > I am very sorry that I have not found time to do much of the > promised > > help. I am in between jobs, and have many things to do, very little > > time. > > I can commit my debian changes if you like, and also do the network > > heavy cvs stuff, but I cannot spend much time on > bugfixing/tracking. > > No problem. We can release .deb once you are ready with it. I will commit the debian stuff to the cvs, 2_6 branch, it just goes into a subdir. > > > > > > > --- Grzegorz Jakacki <ja...@he...> schrieb: > > > > Hi, > > > > > > In anticipation of checkin from Alexandre Tolmos I created > rel_2_6 > > > branch > > > out of current HEAD. This is what is going to end up as 2.6 > release > > > tarball. > > > > > > Also please remember, that today is the last chance to check if > > > > > > > > > > opencxx.sourceforge.net/release-2.6-testing/opencxx-2.6-exp1.tar.gz > > > > I have checked the current cvs 2.6 > > cvs co -r rel_2_6 opencxx > > > > here are the compiler errors : > [...] > > What compiler, what platform? Debian GNU/Linux i386 unstable , gcc 3.2 mike ===== James Michael DuPont http://introspector.sourceforge.net/ __________________________________________________ Do you Yahoo!? Yahoo! Tax Center - forms, calculators, tips, more http://taxes.yahoo.com/ |
|
From: Grzegorz J. <ja...@he...> - 2003-02-21 00:02:21
|
On Thu, 20 Feb 2003, [iso-8859-1] James Michael DuPont wrote: > Gregorz, > I am very sorry that I have not found time to do much of the promised > help. I am in between jobs, and have many things to do, very little > time. > I can commit my debian changes if you like, and also do the network > heavy cvs stuff, but I cannot spend much time on bugfixing/tracking. No problem. We can release .deb once you are ready with it. > > > --- Grzegorz Jakacki <ja...@he...> schrieb: > > > Hi, > > > > In anticipation of checkin from Alexandre Tolmos I created rel_2_6 > > branch > > out of current HEAD. This is what is going to end up as 2.6 release > > tarball. > > > > Also please remember, that today is the last chance to check if > > > > > > opencxx.sourceforge.net/release-2.6-testing/opencxx-2.6-exp1.tar.gz > > I have checked the current cvs 2.6 > cvs co -r rel_2_6 opencxx > > here are the compiler errors : [...] What compiler, what platform? Best regards Grzegorz ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Alexandre T. <kt...@fr...> - 2003-02-20 13:00:49
|
Le jeudi, 20 f=E9v 2003, =E0 02:42 Europe/Paris, Grzegorz Jakacki a =E9cr=
it :
> On Wed, 19 Feb 2003, Alexandre Tolmos wrote:
>
>> Grzegorz,
>>
>> I merged Bruno's work with the current Cvs tree.
>
> Great. (Isn't his name Michael Buro ?)
Yes, sorry :)
>
>> The namespace problem
>> is gone but I'm facing now another problem relating to dynamic libs:
>
> Could you check in what you have before you move on to fixing
> things with Darwin port? It would be better to have two checkins,
> each covering a smaller change, than a big one with everything lumped
> together.
>
> I am attaching checkin instructions at the end of this e-mail. Should
> you encouter any problems, let me know. Also send me you SF user id.
>
>> "metaclass.h" includes some "ltdl.h" header file which does not exist
>> on Mosx (well, Darwin actually).
>
> This is file from libtool and it should be present in =20
> 'opencxx/libltdl/'.
> Is it there?
>
>> Is there a standard way to handle Dlls
>> across multiple Unix variants (Posix?).
>
> Yes. Libtool hides all the plarform specific stuff from you. I found
> "darwin" in several places in libtool script, so it looks like this =20
> system
> is handled.
>
>> Moreover, Dlls have a ".dylib"
>> extension on Darwin, not ".so" like on FreeBSD (although Darwin is Bsd
>> based).
>
> I also fount 'dylib' in libtool sources, so this should be covered =20
> too. If
> you want to give an idea of how libtool works have a look at info docs
> ('info libtool') and Autobook (aka. Goat Book) at
> http://sources.redhat.com/autobook/ .
>
>> Any help is welcome :)
>
> Hope this helps.
Ok. Everything compiles now but all the tests in "testsuite" fail.
Moreover, I have still a problem with "configure" stuff: I have to run =20
"configure" and "make" in the Gc directory first, then run "configure" =20
and "make" in the main directory. Maybe it's because I replaced the old =20
Gc with the latest version (6.1) but kept the original =20
"Makefile.gc.in". I don't know how to deal with those configuration =20
files, I need some help there :)
Anyway, I'll try to checkin the changes made to Occ itself (the Gcc 3 =20
fix).
Alex
------------------------------------------------------------------------=20
-
Alexandre Tolmos
E-mail:=A0...@fr...
ICQ: 92964905
------------------------------------------------------------------------=20
-
"Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn."
------------------------------------------------------------------------=20
-
|
|
From: <mdu...@ya...> - 2003-02-20 12:43:37
|
Gregorz, I am very sorry that I have not found time to do much of the promised help. I am in between jobs, and have many things to do, very little time. I can commit my debian changes if you like, and also do the network heavy cvs stuff, but I cannot spend much time on bugfixing/tracking. --- Grzegorz Jakacki <ja...@he...> schrieb: > > Hi, > > In anticipation of checkin from Alexandre Tolmos I created rel_2_6 > branch > out of current HEAD. This is what is going to end up as 2.6 release > tarball. > > Also please remember, that today is the last chance to check if > > > opencxx.sourceforge.net/release-2.6-testing/opencxx-2.6-exp1.tar.gz I have checked the current cvs 2.6 cvs co -r rel_2_6 opencxx here are the compiler errors : ===== James Michael DuPont http://introspector.sourceforge.net/ __________________________________________________________________ Gesendet von Yahoo! Mail - http://mail.yahoo.de Bis zu 100 MB Speicher bei http://premiummail.yahoo.de |
|
From: Grzegorz J. <ja...@he...> - 2003-02-20 05:13:42
|
Hi,
In anticipation of checkin from Alexandre Tolmos I created rel_2_6 branch
out of current HEAD. This is what is going to end up as 2.6 release
tarball.
Also please remember, that today is the last chance to check if
opencxx.sourceforge.net/release-2.6-testing/opencxx-2.6-exp1.tar.gz
works as expected on your platforms.
Best regards
Grzegorz
##################################################################
# Grzegorz Jakacki Huada Electronic Design #
# Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang #
# tel. +86-10-64365577 x2074 Beijing 100015, China #
# Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. #
##################################################################
|
|
From: Grzegorz J. <ja...@he...> - 2003-02-20 01:44:06
|
On Wed, 19 Feb 2003, Alexandre Tolmos wrote:
> Grzegorz,
>
> I merged Bruno's work with the current Cvs tree.
Great. (Isn't his name Michael Buro ?)
> The namespace problem
> is gone but I'm facing now another problem relating to dynamic libs:
Could you check in what you have before you move on to fixing
things with Darwin port? It would be better to have two checkins,
each covering a smaller change, than a big one with everything lumped
together.
I am attaching checkin instructions at the end of this e-mail. Should
you encouter any problems, let me know. Also send me you SF user id.
> "metaclass.h" includes some "ltdl.h" header file which does not exist
> on Mosx (well, Darwin actually).
This is file from libtool and it should be present in 'opencxx/libltdl/'.
Is it there?
> Is there a standard way to handle Dlls
> across multiple Unix variants (Posix?).
Yes. Libtool hides all the plarform specific stuff from you. I found
"darwin" in several places in libtool script, so it looks like this system
is handled.
> Moreover, Dlls have a ".dylib"
> extension on Darwin, not ".so" like on FreeBSD (although Darwin is Bsd
> based).
I also fount 'dylib' in libtool sources, so this should be covered too. If
you want to give an idea of how libtool works have a look at info docs
('info libtool') and Autobook (aka. Goat Book) at
http://sources.redhat.com/autobook/ .
> Any help is welcome :)
Hope this helps.
Best regards
Grzegorz
--------------------------------------------------------------------
This is draft of something I am planning to put on the website soon.
Sorry for bugs etc., you should get the idea. Contact me
if any issues.
<html>
<body>
<h1>Comitting changes into CVS</h1>
<i>jakacki at users.sourceforge.net</i>
<ul>
<li> Contact project admin to discuss what you want to check in.
<li> Create your account on the SourceForge. Mail your user id to the
project admin.
<li> Make sure you have the proper
<href a=developer-tools-checklist.html>tools</a>.
<li> Read instructions
about <href a=?>setting up CVS developers access</a>
and <href=?>setting up SSH shell access</a> in Site Docs.
<li> Have a look at
<a href=http://cvsbook.red-bean.com/cvsbook.html>CVSBook</a>.
<li> Set:
export CVS_RSH
CVS_RSH=ssh
export CVSROOT
CVSROOT=ext:<i>USERID</i>@cvs.sourceforge.net:/cvsroot/opencxx
<li> Once you are added to the project, create your sandbox
(if admin hasn't do it for you) and check out the sources
from your sandbox:
cvs rtag -b sandbox_<i>SANDBOXNAME</i>
for <i>SANDBOXNAME</i> take <i>USERID</i> (so the tag would be e.g.
'sandbox_jakacki'), optionally followed by "_" and
some text (e.g. 'sandbox_jakacki_refact')
<li> Check out your sandbox:
cvs co -r sandbox_<i>SANDBOXNAME</i> opencxx
<li> Run
cd opencxx
./configure --enable-maintainer-mode
<li> Introduce your changes. If you modify files controlling
the build procedure (e.g. configure.in, Makefile.am etc.), run
./bootstrap
You may also want to consult
<a href=http://sources.redhat.com/autobook>Autobook</a>.
<li> Build
make # gmake on systems with non-GNU make
<li> Test
cd testsuite
make test # or gmake
You should get zero unexpected failures in tests summary.
<li> Commit
cvs commit
You comitted to your sandbox, not to the main trunk.
When asked by cvs, leave the changes description empty.
You will describe the changes when merging your sandbox
into main trunk
<li> Log on to Compile Farm
ssh cf.sourceforge.net
<li> Select the system you want to test on
(from systems at CF we currently support
PC/Linux, Sun/Linux, Sun/Solaris)
<li> Set up CVSROOT and CVS_RSH.
<li> Check out your sandbox:
cvs co -r sandbox_<i>SANDBOXNAME</i> opencxx
<li> Build and test
cd opencxx
./configure
make # or gmake
cd testsuite
make tests # or gmake
You should get zero unexpected failuers in tests summary.
<li> Log out from Compile Farm.
<li> Make your working copy a HEAD revision
cvs update -A
<li> Merge your sandbox into the working copy of HEAD revision.
Look out for CVS reports about conflicts.
cvs update -j sandbox_<i>SANDBOXNAME</i>
<li> If somebody else did checkin while you were working in your
sandbox, you may get conflicts during merge. In such a case:
<ul>
<li> Inspect the files in which conflits were reported and
resolve the conflicts.
<li> Go back to your sandbox with resolved conflicts
cvs update -r sandbox_<i>SANDBOXNAME</i>
<li> Test locally and on Compile Farm.
<li> Try merging again.
</ul>
<li> After you successfully merged your changes, you may want to review
them:
cvs diff
and prepare short descriptions of what your changes do.
<li> Commit
cvs commit
supplying your descriptions. (Set EDITOR to your favorite editor
if you are not satisfied with the one that CVS brings up.)
<li> Done.
</ul>
<br>
[I am working on a script that checkouts, configures, builds
and tests on Compile Farm, so the above is likely to get shorter. GJ]
</body>
</html>
##################################################################
# Grzegorz Jakacki Huada Electronic Design #
# Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang #
# tel. +86-10-64365577 x2074 Beijing 100015, China #
# Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. #
##################################################################
|
|
From: Alexandre T. <kt...@fr...> - 2003-02-19 21:04:05
|
>>>> I replaced the current Gc in the Cvs version with Gc 6.1. It =20 >>>> compiles >>>> without any problem but Occ itself does not compile at all, I get =20 >>>> tons >>>> of errors. The Gcc version >>>> shipped with the Mosx dev package is 3.1. Which version do you use =20 >>>> to >>>> compile Occ? >>> >>> OpenC++ sources does not conform to namespaces scheme mandated by C++ >>> standard >>> (i.e. that standard library components live in 'std' namespace). >>> OpenC++ >>> DOES NOT compile with gcc>=3D3.0. It is not difficult to fix and >>> somebody has already done it, but nobody had enough will to check it >>> in. >>> Go to OpenC++ project page at www.sf.net and search the archive of >>> opencxx-user, the link to the gcc-3.0 compatible version has been >>> posted >>> here. >> >> I'll try that. > > I am very sorry. I checked myself and it is not there (we started > archiving messages later than I thought). This is what you need: > > http://www.neci.nj.nec.com/homepages/mic/openc++2.5.12-gcc-3.0.3.tgz > > The fix was prepared by Michael Buro (mic at research dot nj dot nec > dot com). Again, sorry for that. > >> >>> If you can spare some time to merge the changes with HEAD revision >>> and check it in, the heavens will thank you (and I will join too). >> >> No problem but I haven't used CVS for quite a while... > > I can help if there are any issues. > >>>> Note: to compile Occ, I duplicated "Makefile.FreeBSD" in "src/Unix" >>>> and >>>> used "make -f Makefile.Darwin all". >>> >>> *PLEASE* work on CVS HEAD. >>> >>> 2.5.12/2.5.13 is difficult to maintain because of code duplication in >>> its >>> build system. OpenC++ project does not have resources to maintain it. >>> Moreover, 2.5.12 does not include test harness nor test suite, which >>> disables reliable development, because you have no means to verify if >>> your >>> contribution breaks the current functionality. It is likely that your >>> otherwise very valuable work will not be useful if you stick to =20 >>> 2.5.12 >>> . >> >> Don't worry, I'm working on the latest tree posted on SourceForge :) > > Cool. Grzegorz, I merged Bruno's work with the current Cvs tree. The namespace problem =20 is gone but I'm facing now another problem relating to dynamic libs: =20 "metaclass.h" includes some "ltdl.h" header file which does not exist =20 on Mosx (well, Darwin actually). Is there a standard way to handle Dlls =20 across multiple Unix variants (Posix?). Moreover, Dlls have a ".dylib" =20 extension on Darwin, not ".so" like on FreeBSD (although Darwin is Bsd =20 based). Any help is welcome :) Alex ------------------------------------------------------------------------=20 - Alexandre Tolmos E-mail:=A0...@fr... ICQ: 92964905 ------------------------------------------------------------------------=20 - "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn." ------------------------------------------------------------------------=20 - |
|
From: Alexandre T. <kt...@fr...> - 2003-02-19 16:31:26
|
Le mercredi, 19 f=E9v 2003, =E0 01:40 Europe/Paris, Grzegorz Jakacki a =20 =E9crit : > On Tue, 18 Feb 2003, Alexandre Tolmos wrote: > >> >> Le mardi, 18 f=E9v 2003, =E0 02:22 Europe/Paris, Grzegorz Jakacki a =E9= crit =20 >> : >> >>>> I replaced the current Gc in the Cvs version with Gc 6.1. It =20 >>>> compiles >>>> without any problem but Occ itself does not compile at all, I get =20 >>>> tons >>>> of errors. The Gcc version >>>> shipped with the Mosx dev package is 3.1. Which version do you use =20 >>>> to >>>> compile Occ? >>> >>> OpenC++ sources does not conform to namespaces scheme mandated by C++ >>> standard >>> (i.e. that standard library components live in 'std' namespace). >>> OpenC++ >>> DOES NOT compile with gcc>=3D3.0. It is not difficult to fix and >>> somebody has already done it, but nobody had enough will to check it >>> in. >>> Go to OpenC++ project page at www.sf.net and search the archive of >>> opencxx-user, the link to the gcc-3.0 compatible version has been >>> posted >>> here. >> >> I'll try that. > > I am very sorry. I checked myself and it is not there (we started > archiving messages later than I thought). This is what you need: > > http://www.neci.nj.nec.com/homepages/mic/openc++2.5.12-gcc-3.0.3.tgz > > The fix was prepared by Michael Buro (mic at research dot nj dot nec > dot com). Again, sorry for that. > Actually this link was broken too :) The new link is: =20 http://www.cs.ualberta.ca/~mburo/openc++2.5.12-gcc-3.0.3.tgz I'm trying that Asap. Thanks. Alex ------------------------------------------------------------------------=20 - Alexandre Tolmos E-mail:=A0...@fr... ICQ: 92964905 ------------------------------------------------------------------------=20 - "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn." ------------------------------------------------------------------------=20 - |
|
From: Grzegorz J. <ja...@he...> - 2003-02-19 00:42:54
|
On Tue, 18 Feb 2003, Alexandre Tolmos wrote: > > Le mardi, 18 f=E9v 2003, =E0 02:22 Europe/Paris, Grzegorz Jakacki a =E9= crit : > > >> I replaced the current Gc in the Cvs version with Gc 6.1. It compile= s > >> without any problem but Occ itself does not compile at all, I get to= ns > >> of errors. The Gcc version > >> shipped with the Mosx dev package is 3.1. Which version do you use t= o > >> compile Occ? > > > > OpenC++ sources does not conform to namespaces scheme mandated by C++ > > standard > > (i.e. that standard library components live in 'std' namespace). > > OpenC++ > > DOES NOT compile with gcc>=3D3.0. It is not difficult to fix and > > somebody has already done it, but nobody had enough will to check it > > in. > > Go to OpenC++ project page at www.sf.net and search the archive of > > opencxx-user, the link to the gcc-3.0 compatible version has been > > posted > > here. > > I'll try that. I am very sorry. I checked myself and it is not there (we started archiving messages later than I thought). This is what you need: http://www.neci.nj.nec.com/homepages/mic/openc++2.5.12-gcc-3.0.3.tgz The fix was prepared by Michael Buro (mic at research dot nj dot nec dot com). Again, sorry for that. > > > If you can spare some time to merge the changes with HEAD revision > > and check it in, the heavens will thank you (and I will join too). > > No problem but I haven't used CVS for quite a while... I can help if there are any issues. > >> Note: to compile Occ, I duplicated "Makefile.FreeBSD" in "src/Unix" > >> and > >> used "make -f Makefile.Darwin all". > > > > *PLEASE* work on CVS HEAD. > > > > 2.5.12/2.5.13 is difficult to maintain because of code duplication in > > its > > build system. OpenC++ project does not have resources to maintain it. > > Moreover, 2.5.12 does not include test harness nor test suite, which > > disables reliable development, because you have no means to verify if > > your > > contribution breaks the current functionality. It is likely that your > > otherwise very valuable work will not be useful if you stick to 2.5.1= 2 > > . > > Don't worry, I'm working on the latest tree posted on SourceForge :) Cool. Regards Grzegorz ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Alexandre T. <kt...@fr...> - 2003-02-18 22:32:52
|
Le mardi, 18 f=E9v 2003, =E0 02:22 Europe/Paris, Grzegorz Jakacki a =E9cr= it : >> I replaced the current Gc in the Cvs version with Gc 6.1. It compiles >> without any problem but Occ itself does not compile at all, I get tons >> of errors. The Gcc version >> shipped with the Mosx dev package is 3.1. Which version do you use to >> compile Occ? > > OpenC++ sources does not conform to namespaces scheme mandated by C++ =20 > standard > (i.e. that standard library components live in 'std' namespace). =20 > OpenC++ > DOES NOT compile with gcc>=3D3.0. It is not difficult to fix and > somebody has already done it, but nobody had enough will to check it =20 > in. > Go to OpenC++ project page at www.sf.net and search the archive of > opencxx-user, the link to the gcc-3.0 compatible version has been =20 > posted > here. I'll try that. > If you can spare some time to merge the changes with HEAD revision > and check it in, the heavens will thank you (and I will join too). No problem but I haven't used CVS for quite a while... >> Note: to compile Occ, I duplicated "Makefile.FreeBSD" in "src/Unix" =20 >> and >> used "make -f Makefile.Darwin all". > > *PLEASE* work on CVS HEAD. > > 2.5.12/2.5.13 is difficult to maintain because of code duplication in =20 > its > build system. OpenC++ project does not have resources to maintain it. > Moreover, 2.5.12 does not include test harness nor test suite, which > disables reliable development, because you have no means to verify if =20 > your > contribution breaks the current functionality. It is likely that your > otherwise very valuable work will not be useful if you stick to 2.5.12 = =20 > . Don't worry, I'm working on the latest tree posted on SourceForge :) Thanks for your help. Regards, Alex ------------------------------------------------------------------------=20 - Alexandre Tolmos E-mail:=A0...@fr... ICQ: 92964905 ------------------------------------------------------------------------=20 - "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn." ------------------------------------------------------------------------=20 - |
|
From: Shigeru C. <ch...@is...> - 2003-02-18 14:17:01
|
From: Grzegorz Jakacki <ja...@he...> Date: Tue, 18 Feb 2003 14:30:58 +0800 (CST) > > > > Memory management with smart pointers > > > > should be slower than with GC... > > > > > > Why? I do not disagree, but I am still looking for convincing argument. > > > > Yes, the performance depends on memory usage patterns. > > > > GC should work better if a large number of objects are allocated and > > soon deallocated before the next GC phase. Suppose that 100 objects > > are allocated and then 90 of them are deallocated. The deallocation > > cost with smart pointers is 90 * <time for free()>. On the other hand, > > if GC (except reference counting GC) is used, the deallocation cost > > is 10 * <time for collecting a live object>. > > Doesn't GC has to free those 90 dead (leaked) objects anyway? Oops, the example above is for a copying GC (or any other modern GC) but conservative GC is mark-sweep GC (as far as I know). If GC is a copying GC (or any other modern GC), the whole memory region including the dead objects is deallocated after live objects are taken from that region to another region. So the deallocation cost is in principle independent of the number of dead objects. It depends on the number of live objects. But I still think conservative GC (mark-sweep GC) is mostly better than smart pointers, which is based on reference counting GC. For example, creating and deleting smart pointers includes runtime penalties due to maintaining reference counts. If mark-sweep GC is used, such a house-keeping job is done only at GC time. Forthermore, smart pointers introduce extra indirection to every access. Chiba |
|
From: Grzegorz J. <ja...@he...> - 2003-02-18 11:56:24
|
On Tue, 18 Feb 2003, grigoriev wrote: > Hello, > > Problem - is in wide application of Open C++, and making this practical > one. > And making real programmers having true good tool.(!) > this means that commercial applications often is developed by > Microsoft Visual C++ 7 compatible tons of code, and not abstract ones. > > Asked about AOP: > Aspect Oriented Programming - a kind of meta programming - > (analog declaration modifiers of Open C++) > is based on propose that to the base language constructions may be added > some declararative type > features (for characterizing base language constructions, to be coding units > self) > that can be used for productive realizing of many design patterns - > and this is often most efficient way to deal with comprehensive > metaprogramming. > (Can not be realized by structures of usual language) > In Open C++ this is done by declaration modifiers, and have simple structure > (syntax) compared to > Microsoft version of AOP. > In MS it is possible to define a new ones -with arguments(!!!), some good > syntax with structure of checking their type safe at compile time. > Does it work so at open C++? No, currently OpenC++ does not implement arguments to declaration modifiers. > Can be Open C++ extended or modified or make be more better and wide? And > useful? There is declared ways? Or just "as is"? OpenC++ can be extended to be whatever you want it to be, provided that you do it by yourself or find resources to do it. OpenC++ is maintained by a group of volunteers who are busy earning their salaries or university degrees and usually can spare only a little time for OpenC++. > > Can I make myself analog of [attribute() modifier] syntax to make parsed > Microsoft code by Open C++ - > that means self-metaprogramming in OpenC++. Technically it should be possible. However we do not want introduce changes that break backward compatibility (if the feature you are requesting has to do it, it perhaps should be activated only on user's demand). > For: > Again, > For real work and application of meta programming It is better to make Open > C++ > compatible with libraries often used in industry. > Does it hard to make Open C++ compatible with Microsoft version ? Probably not. Depends on what you mean by "hard". I could perhaps give more accurate answer if you supply e.g. examples of what you would like in OpenC++. > Instead of writing just generation of this code self. > Parsing of existing code and than meta refactoring- is better!! -The purpose > of my ask question. > > -- To make effective refactored existing code for really good use. - tons > of goods.!!-The second real purpose. > > It is better to have similar coding in Meta Compiling and base language > modifiers on OpenC++ is done without [], and extended syntax. > (Can be looked better at C# documentation). OpenC++ was implemented many years before C# was designed. > Do you abstract programmers (for selves) or real - for practice ones ? > finally. What is "abstract programmer"? Best regards Grzegorz > > > thanks you for you response. > > > Vladimir Grigoriev. > > I'm working in Microsoft Visual C++ 7, > > > > > > and some compatibility problem there is: > > > > > > Does it fully compatible with Microsoft Visual C++ 7? > > > Especially -With Attributing programming - new constructions added. > > > > I do not think so. Can you expand on how this attributing is supposed to > > work? Perhaps we can figure out how far from that the OpenC++ is. > > > > > Does it fully compatible with Microsoft Visual C++ 6? -Standard > > > compatibility. > > > > I do not use MSVC++, I do not know what this switch really does. Perhaps > > you want to elaborate on this or wait for somebody else to address this > > question. > > > > > "[]" the square brackets constructions possible to parse in Open C++ > > > by -self meta programming? > > > > It is not quite clear to me what you mean here. AFAIR OpenC++ parses > > square bracket initializers. What is "self meta programming"? > > > > > Can I parse and meta-use that new technique of expressing AOP ? > > > AOP - is Some core technology of meta programming - subject of Open C++. > > > > Could you explain what AOP is? > > > > > Or can I just parse remaining code with attributes presented in it? > > > > This question is not clear to me either. > > > > Best regards > > Grzegorz > > > > ################################################################## > > # Grzegorz Jakacki Huada Electronic Design # > > # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # > > # tel. +86-10-64365577 x2074 Beijing 100015, China # > > # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # > > ################################################################## > > > > > > > ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: grigoriev <gri...@ar...> - 2003-02-18 09:58:39
|
Hello, Problem - is in wide application of Open C++, and making this practical one. And making real programmers having true good tool.(!) this means that commercial applications often is developed by Microsoft Visual C++ 7 compatible tons of code, and not abstract ones. Asked about AOP: Aspect Oriented Programming - a kind of meta programming - (analog declaration modifiers of Open C++) is based on propose that to the base language constructions may be added some declararative type features (for characterizing base language constructions, to be coding units self) that can be used for productive realizing of many design patterns - and this is often most efficient way to deal with comprehensive metaprogramming. (Can not be realized by structures of usual language) In Open C++ this is done by declaration modifiers, and have simple structure (syntax) compared to Microsoft version of AOP. In MS it is possible to define a new ones -with arguments(!!!), some good syntax with structure of checking their type safe at compile time. Does it work so at open C++? Can be Open C++ extended or modified or make be more better and wide? And useful? There is declared ways? Or just "as is"? Can I make myself analog of [attribute() modifier] syntax to make parsed Microsoft code by Open C++ - that means self-metaprogramming in OpenC++. For: Again, For real work and application of meta programming It is better to make Open C++ compatible with libraries often used in industry. Does it hard to make Open C++ compatible with Microsoft version ? Instead of writing just generation of this code self. Parsing of existing code and than meta refactoring- is better!! -The purpose of my ask question. -- To make effective refactored existing code for really good use. - tons of goods.!!-The second real purpose. It is better to have similar coding in Meta Compiling and base language modifiers on OpenC++ is done without [], and extended syntax. (Can be looked better at C# documentation). Do you abstract programmers (for selves) or real - for practice ones ? finally. thanks you for you response. Vladimir Grigoriev. I'm working in Microsoft Visual C++ 7, > > > > and some compatibility problem there is: > > > > Does it fully compatible with Microsoft Visual C++ 7? > > Especially -With Attributing programming - new constructions added. > > I do not think so. Can you expand on how this attributing is supposed to > work? Perhaps we can figure out how far from that the OpenC++ is. > > > Does it fully compatible with Microsoft Visual C++ 6? -Standard > > compatibility. > > I do not use MSVC++, I do not know what this switch really does. Perhaps > you want to elaborate on this or wait for somebody else to address this > question. > > > "[]" the square brackets constructions possible to parse in Open C++ > > by -self meta programming? > > It is not quite clear to me what you mean here. AFAIR OpenC++ parses > square bracket initializers. What is "self meta programming"? > > > Can I parse and meta-use that new technique of expressing AOP ? > > AOP - is Some core technology of meta programming - subject of Open C++. > > Could you explain what AOP is? > > > Or can I just parse remaining code with attributes presented in it? > > This question is not clear to me either. > > Best regards > Grzegorz > > ################################################################## > # Grzegorz Jakacki Huada Electronic Design # > # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # > # tel. +86-10-64365577 x2074 Beijing 100015, China # > # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # > ################################################################## > > |
|
From: Grzegorz J. <ja...@he...> - 2003-02-18 06:34:25
|
On Tue, 18 Feb 2003, Shigeru Chiba wrote: > > From: Grzegorz Jakacki <ja...@he...> > Subject: [Opencxx-users] Re: [opencxx] Porting to MacOs X > Date: Tue, 18 Feb 2003 08:48:21 +0800 (CST) > > > > > > PS: If this does not work, however, there is possibly a way to get rid > > > > > of GC at all with help of boost::shared_ptr<>. > > > > > > Memory management with smart pointers > > > should be slower than with GC... > > > > Why? I do not disagree, but I am still looking for convincing argument. > > Yes, the performance depends on memory usage patterns. > > GC should work better if a large number of objects are allocated and > soon deallocated before the next GC phase. Suppose that 100 objects > are allocated and then 90 of them are deallocated. The deallocation > cost with smart pointers is 90 * <time for free()>. On the other hand, > if GC (except reference counting GC) is used, the deallocation cost > is 10 * <time for collecting a live object>. Doesn't GC has to free those 90 dead (leaked) objects anyway? Regards Grzegorz > I know this discussion is very rough, but my understanding is GC is > rather fast under some circumstances. > > Chiba > > ################################################################## # Grzegorz Jakacki Huada Electronic Design # # Senior Engineer, CAD Dept. 1 Gaojiayuan, Chaoyang # # tel. +86-10-64365577 x2074 Beijing 100015, China # # Copyright (C) 2002 Grzegorz Jakacki, HED. All Rights Reserved. # ################################################################## |
|
From: Shigeru C. <ch...@is...> - 2003-02-18 05:42:07
|
From: Grzegorz Jakacki <ja...@he...> Subject: [Opencxx-users] Re: [opencxx] Porting to MacOs X Date: Tue, 18 Feb 2003 08:48:21 +0800 (CST) > > > > PS: If this does not work, however, there is possibly a way to get rid > > > > of GC at all with help of boost::shared_ptr<>. > > > > Memory management with smart pointers > > should be slower than with GC... > > Why? I do not disagree, but I am still looking for convincing argument. Yes, the performance depends on memory usage patterns. GC should work better if a large number of objects are allocated and soon deallocated before the next GC phase. Suppose that 100 objects are allocated and then 90 of them are deallocated. The deallocation cost with smart pointers is 90 * <time for free()>. On the other hand, if GC (except reference counting GC) is used, the deallocation cost is 10 * <time for collecting a live object>. I know this discussion is very rough, but my understanding is GC is rather fast under some circumstances. Chiba |
7 messages has been excluded from this view by a project administrator.
