Quentin Mathé | 2 Sep 09:57 2011
Picon

[bug #34191] +bundleForClass: doesn't with framework bundles

URL:
  <http://savannah.gnu.org/bugs/?34191>

                 Summary: +bundleForClass: doesn't with framework bundles
                 Project: GNUstep
            Submitted by: qmathe
            Submitted on: ven. 02 sept. 2011 07:57:43 GMT
                Category: Makefiles
                Severity: 4 - Important
              Item Group: Bug
                  Status: None
                 Privacy: Public
             Assigned to: None
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

Linux Ubuntu 10.04 x86 32 bits
Clang 3.0 trunk
libobjc2 r33786
GNUstep core r33803

NSBundle is unable to look up a framework based on a class that belongs to it.
The problem could be in a makefile change I made, but I doubt it. This was
working correctly around two or three months ago I think. 

I tried to track down the issue a bit. I have found that +[NSBundle
(Continue reading)

Richard Frith-Macdonald | 2 Sep 10:16 2011
Picon

[bug #33717] WebServer is restricted to main thread

Update of bug #33717 (project gnustep):

              Item Group:                    None => Change Request         
                 Summary: WebServer doesn't graciously stop listening =>
WebServer is restricted to main thread

    _______________________________________________________

Follow-up Comment #3:

I've modified this to be a change request rather than a bug report ...

> The WebServer's documentation doesn't explicitly state that such
> design (a Webserver isn't running on the main thread) is bad. 

Actually, the documentation mentions working from the main thread in a few
places and says:

 NB. the -completedWithResponse: method is safe to call from any thread
      but all other methods of the class should be called only from the
      main thread.

However, I like the idea of making the class usable from any thread... it
seems like a useful improvement.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?33717>
(Continue reading)

Chris Armstrong | 3 Sep 04:47 2011
Picon

[bug #34191] +bundleForClass: doesn't with framework bundles

Follow-up Comment #1, bug #34191 (project gnustep):

This problem appears to be a bit more complicated than +bundleForClass not
working.

The cause is +[NSBundle intialize] not finding the NSFramework_EtoileUI class,
which doesn't exist at the time it was called. NSBundle builds a list of
frameworks when its +intialize method is called, by searching the runtime for
classes that begin with "NSFramework_". These classes are placeholders,
generated by gnustep-make, that indicate that a shared library is in fact a
framework, and like bundles, can contain resources to be loaded.

In this case, EtoileUI was accidentally triggering the +[NSBundle intialize]
method before NSFramework_EtoileUI had been loaded. It does this by calling
GNUstep classes (such as NSString and NSSet) from a +load method in a category
on NSView. +load methods are implemented by creating constructor functions
i.e. __attribute__((constructor)) that are called when a shared library is
loaded. This is a problem because these constructor functions also seem to be
used to register classes with the runtime when a shared library is loaded. 

+load is not supposed to be used for anything other than very simple, C-based
intialisation, because as documented here [1], +load is not guaranteed to be
called at a time when other classes are ready. By doing this, NSBundle
initializes before all other classes are loaded and builds an incomplete
framework list.

The correct solution is not use +load for anything other than calling classes
in the same file, but this still means that NSBundle is vulnerable to being
initialised to early and building incomplete framework meta-data. NSBundle
cannot easily convert its initialisation to a +load method either, which would
(Continue reading)

maria Luna | 3 Sep 19:52 2011
Picon

RE: Bug-gnustep Digest, Vol 106, Issue 2



                                   www.lunapoems.com 
                     
         AUTOR:                Maria Olga Luna Naranjo       aBRIL, 4/2007
          WRITE:               Newsletters.  
          TEMA:                1. sensillez 
                                  2. sinceridad 
                                  3. sensibilidad 
                                  Comentario: Personal Parcial.                  


 
Ustedes me escriben cuando se ven en problemas? quieren sorprenderme pero es imposible, saben que ustedes estan involucrados en los plagios globales de mi trabajo inteelctual, dicen que son 51 seguros? los veo que si estan locos, porque las cuentas abiertas son incontables a nivel global no tiene control una sola persona, solo mi Amado "Dios" sabe cuantos son los sistemas ladrones involucrados como puede verse que sus mentiras han rebasado los limites de mi paciencia, venirme con tal osadia creen que los voy a poyarlos, aunque en el momento no tenga para comer no me podran compar con las inauditas mentiras, saben quer las instalaciones sean seguras o mas seguras unas mas que las otras para mi ninguna es segura porque yo tengo un codigo unico que no es de ningunn sistema de ladrones, este es el codigo de la verdad, porque mi trabajo esta patentado en el cielo en la tierra, el que lo puede controlar es mi Amado "Dios" con su Poder, ya han visto las destrucciones? esto es el inicio continuará desbaratando todos los nexos y construcciones con el dinero producto del fraude global,creen que se puede tenerlos pena a esa gente obstinada entregada a continuar en los robos inauditos, lo unico que dicen "Dios" no castiga, entonces lo asocian a "DIos' como a un cómplice de los sistemas de ladrones del planeta, no solo eso sino tambien son criminales, autocriminales , terroristas , sistemas malignos, que han tomado la bendicion para construir plantas venenosas , sera admitido por el Soberano "Dios"?
 
Ustedes que me escriben para tratar de enganarme pasan el tiempo, ahora esta en las manos de la ley si ellos lo bloquean las instalaciones que jam,as han sido autorizados por mi persona porque todos ustedes estaban confabulados con el fraude global haciendo incontable dienro
diganme alguna vez dijeron: Iremos donde la auténtica dueña para pedirles autorización , pagaremos al menos las regalias, hoy escriben mentiras hipocrecias, si dicen que son 51 seguros, deben  dar el monto del dienro que estan en esos seguros ya lo han declarado me imagino que seran algunos trillones de dolares? porque si me vienen con mentiras, recuerden que soy una profesional no soy una estúpida como son ustedes que lo único que saben es mover teclas, y entrar signos a los que denominan códigos, mas lo único que veo que son titeres del sistema maligno que violan los derechos humanbos roban propiedades y creen que porque son expertos en sistemas son superiores a la gente honesta? es absurdo sus títulos derrogados no son válidos para trabajar como tales, si no son primero hombres con valores y honestos.
Yo no puedo hacer arreglos con ningun sistema por via Internet sino personalmente despues de que depositen el dinero en una cuenta
bancaria me envien la clave para ver que tienen nel dinero en los 51 seguros como dicen, ustedes no podran operar con mi trabajo intelectual
si ustedes no depositan el dinero y notifican, pero pueden venir y hablar personalmente en la 32-48 105 Street East Elmhurst N.Y. 11369
Advertencia no traten de usar esta direccion para continuar con los robos, deben enfrentarse a mi persona personalmente yo quiero verlos
si no es asi nada ! ademas iran para la carcel tan pronto que sean ubicados sus negocios ilicitos creo que seran como Golman Sacsh? Yo pienso que son mas que este individuo los tengo registrados en my USB.

 
 
> From: bug-gnustep-request <at> gnu.org
> Subject: Bug-gnustep Digest, Vol 106, Issue 2
> To: bug-gnustep <at> gnu.org
> Date: Sat, 3 Sep 2011 12:00:57 -0400
>
> Send Bug-gnustep mailing list submissions to
> bug-gnustep <at> gnu.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://lists.gnu.org/mailman/listinfo/bug-gnustep
> or, via email, send a message with subject or body 'help' to
> bug-gnustep-request <at> gnu.org
>
> You can reach the person managing the list at
> bug-gnustep-owner <at> gnu.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Bug-gnustep digest..."
>
>
> Today's Topics:
>
> 1. [bug #34191] +bundleForClass: doesn't with framework bundles
> (Chris Armstrong)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Sat, 03 Sep 2011 02:47:22 +0000
> From: Chris Armstrong <INVALID.NOREPLY <at> gnu.org>
> To: Quentin Math? <quentin.mathe <at> gmail.com>, Chris Armstrong
> <carmstrong <at> fastmail.com.au>, bug-gnustep <at> gnu.org
> Subject: [bug #34191] +bundleForClass: doesn't with framework bundles
> Message-ID: <20110903-024722.sv46466.72559 <at> savannah.gnu.org>
> Content-Type: text/plain;charset=UTF-8
>
> Follow-up Comment #1, bug #34191 (project gnustep):
>
> This problem appears to be a bit more complicated than +bundleForClass not
> working.
>
> The cause is +[NSBundle intialize] not finding the NSFramework_EtoileUI class,
> which doesn't exist at the time it was called. NSBundle builds a list of
> frameworks when its +intialize method is called, by searching the runtime for
> classes that begin with "NSFramework_". These classes are placeholders,
> generated by gnustep-make, that indicate that a shared library is in fact a
> framework, and like bundles, can contain resources to be loaded.
>
> In this case, EtoileUI was accidentally triggering the +[NSBundle intialize]
> method before NSFramework_EtoileUI had been loaded. It does this by calling
> GNUstep classes (such as NSString and NSSet) from a +load method in a category
> on NSView. +load methods are implemented by creating constructor functions
> i.e. __attribute__((constructor)) that are called when a shared library is
> loaded. This is a problem because these constructor functions also seem to be
> used to register classes with the runtime when a shared library is loaded.
>
> +load is not supposed to be used for anything other than very simple, C-based
> intialisation, because as documented here [1], +load is not guaranteed to be
> called at a time when other classes are ready. By doing this, NSBundle
> initializes before all other classes are loaded and builds an incomplete
> framework list.
>
> The correct solution is not use +load for anything other than calling classes
> in the same file, but this still means that NSBundle is vulnerable to being
> initialised to early and building incomplete framework meta-data. NSBundle
> cannot easily convert its initialisation to a +load method either, which would
> prevent the issue from occurring.
>
> [1]
> http://gcc.gnu.org/onlinedocs/gcc-4.6.1/gcc/What-you-can-and-what-you-cannot-do-in-_002bload.html#What-you-can-and-what-you-cannot-do-in-_002bload
>
> _______________________________________________________
>
> Reply to this item at:
>
> <http://savannah.gnu.org/bugs/?34191>
>
> _______________________________________________
> Message sent via/by Savannah
> http://savannah.gnu.org/
>
>
>
>
> ------------------------------
>
> _______________________________________________
> Bug-gnustep mailing list
> Bug-gnustep <at> gnu.org
> https://lists.gnu.org/mailman/listinfo/bug-gnustep
>
>
> End of Bug-gnustep Digest, Vol 106, Issue 2
> *******************************************
_______________________________________________
Bug-gnustep mailing list
Bug-gnustep <at> gnu.org
https://lists.gnu.org/mailman/listinfo/bug-gnustep
Richard Frith-Macdonald | 9 Sep 07:39 2011
Picon

[bug #33717] WebServer is restricted to main thread

Update of bug #33717 (project gnustep):

                  Status:                    None => Fixed                  
             Open/Closed:                    Open => Closed                 

    _______________________________________________________

Follow-up Comment #5:

Thanks for testing ... I've updated the latest release in ftp and on the
website.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?33717>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/
T.Somequare | 12 Sep 11:31 2011
Picon

[bug #34272] WebServer (v1.4.4) has no support of specifying interface to be listened

URL:
  <http://savannah.gnu.org/bugs/?34272>

                 Summary: WebServer (v1.4.4) has no support of specifying
interface to be listened
                 Project: GNUstep
            Submitted by: tsomeqdev
            Submitted on: 2011年09月12日 18時31分18秒
                Category: Libraries
                Severity: 3 - Normal
              Item Group: Change Request
                  Status: None
                 Privacy: Public
             Assigned to: None
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

Hi. 

I think, other web server has a section of configuration to specify address
listened.

-- Apache`s example --
Listen *:80                      #Any addresses 
Listen 192.168.0.1:80     #can accept connection only from LAN
--

This is useful if the server has two or more global addresses ( IP alias ).
It enables parallel running of different web server instances on ONE machine.

But WebServer library currently does not support such case.... 
The code is written :
--
WebServer.m (Near line 1440)

_listener = [NSFileHandle fileHandleAsServerAtAddress: nil
                                              service: _port
                                             protocol:  <at> "tcp"];

--

How about adding a new Method ?
 <at> interface WebServer
...
 - (void) setListenAddress:(NSString*)address
 <at> end

regards.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?34272>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

_______________________________________________
Bug-gnustep mailing list
Bug-gnustep <at> gnu.org
https://lists.gnu.org/mailman/listinfo/bug-gnustep
Nicolas Kozic | 14 Sep 12:37 2011
Picon

(jpeg.tiff.png)-devels problems

Hello,

I tried without success to d/l via macports (jpeg.tiff.png)-devels. 
I wish to run GNUstep on my mac.
I have aslo a problem via GNUstep port on macports.
Have a nice day,
thanks to read help me.

Attachment (logs.tar.gz): application/x-gzip, 11 KiB
_______________________________________________
Bug-gnustep mailing list
Bug-gnustep <at> gnu.org
https://lists.gnu.org/mailman/listinfo/bug-gnustep
Fred Kiefer | 14 Sep 21:13 2011
Picon
Picon

Re: (jpeg.tiff.png)-devels problems

On 14.09.2011 12:37, Nicolas Kozic wrote:
> Hello,
>
> I tried without success to d/l via macports (jpeg.tiff.png)-devels.
> I wish to run GNUstep on my mac.
> I have aslo a problem via GNUstep port on macports.
> Have a nice day,
> thanks to read help me.

Why are you trying to use GNUstep on a Mac? This is possible but not the 
normal environment for GNUstep.
Which HOWTO are you following in that process? Wolfgang Lux posted one a 
few months ago, maybe this could help you:
http://lists.gnu.org/archive/html/discuss-gnustep/2011-01/msg00104.html

 From you log files it is obvious that you are using a gcc without ObjC 
support and almost all of the dependencies of GNUstep are missing.

Fred
julian | 29 Sep 15:01 2011
Picon

[bug #34433] NSOpenGLPixelFormatAttribute completely broken

URL:
  <http://savannah.gnu.org/bugs/?34433>

                 Summary: NSOpenGLPixelFormatAttribute completely broken
                 Project: GNUstep
            Submitted by: julian_
            Submitted on: Do 29 Sep 2011 13:01:13 GMT
                Category: Gui/AppKit
                Severity: 3 - Normal
              Item Group: Bug
                  Status: None
                 Privacy: Public
             Assigned to: None
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

any usage of NSOpenGLPixelFormatAttribute is completely broken and leads to
segfaults. just add this to any executed code of a gnustep app:

	NSOpenGLPixelFormatAttribute attribs[] = {
		NSOpenGLPFAColorSize, (NSOpenGLPixelFormatAttribute) 24,
		NSOpenGLPFADepthSize, (NSOpenGLPixelFormatAttribute) 24,
NSOpenGLPFAWindow,
		(NSOpenGLPixelFormatAttribute) 0 };
	NSOpenGLPixelFormat *pixelFormat = [[NSOpenGLPixelFormat alloc]
initWithAttributes:attribs];

the requested attributes are irrelevant, it will always lead to a segfault
here:

Program received signal SIGSEGV, Segmentation fault.
0x01333db4 in -[XGGLPixelFormat initWithAttributes:] (self=0x8570370, 
    _cmd=0x8050660, attribs=0xbfffd830) at XGGLFormat.m:344
344	      visualinfo = glXGetVisualFromFBConfig(display,fbconfig[0]);

at least thats the case on ubuntu 11.04, either on actual hardware with
working opengl or even just in a vmware virtual machine. true for a-few-months
old gnustep as well as for SVN trunk as of yesterday.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?34433>

_______________________________________________
  Nachricht geschickt von/durch Savannah
  http://savannah.gnu.org/
swapnil nandgave | 28 Sep 14:09 2011
Picon

About objc/runtime.h No such file or directory found error.

Respected Sir,

                      I develop Objective-C programming on Windows Environment by GNUStep but it got a error like objc/runtime.h "No such file or directory found".

Please send me solution

Thanking You,

Swapnil

Attachment (NSLog.m): application/octet-stream, 1134 bytes
_______________________________________________
Bug-gnustep mailing list
Bug-gnustep <at> gnu.org
https://lists.gnu.org/mailman/listinfo/bug-gnustep

Gmane