Re: diffs for NSData.m and Unicode.m (gnustep-base)

I added the NSData fix. The Unicode.m fix doesn't seem to help on my 
system, so I suspect it will always be a problem.

--

-- 
Adam Fedor, Digital Optics            | Fudd's law of opposition: Push
fedor <at> doc.com  http://www.doc.com     | something hard enough, and it
fedor <at> gnu.org  http://www.gnustep.org | will fall over.

NSTextView: -setTextContainerInset patch (take 2)

Hi,

I've re-written my previous patches since they were causing some little 
rendering glitches.

Here's a simple screenshot of Ink.app with a 20,20 inset.

http://sophos.ca/crap/ink.gif

I will continue to work on the text related classes in order to make them 
work better. I plan to work on 'dead keys' support, on 'carriage return' 
problems and other stuff.

Thanks a lot,

Ludo

--

-- 
Live as if you were to die tomorrow.
Learn as if you were to live forever.
 - Gandhi

--- /tmp/gui/Source/GSSimpleLayoutManager.m	Sun Sep 23 18:09:48 2001
+++ GSSimpleLayoutManager.m	Thu Nov  1 12:31:49 2001
 <at>  <at>  -273,7 +273,13  <at>  <at> 
 
   if (![_textStorage length] || ![_lineLayoutInformation count])
     {

Re: diffs for NSData.m and Unicode.m (gnustep-base)

On Thu, 01 Nov 2001 08:42:51 -0700
Adam Fedor <fedor <at> doc.com> wrote:
> I added the NSData fix.

Ah, nice.

> The Unicode.m fix doesn't seem to help on my system, so I suspect
> it will always be a problem.

Hmm, I'm using giconv instead of the plain/original iconv. Could be
due to that.

--

-- 
Chris

Gorm bugs (with patch)

Hi,

In GormDocument: -dealloc we have:

- (void) dealloc
{
  [[NSNotificationCenter defaultCenter] removeObserver: self];
  [window setDelegate: nil];
  [window performClose: self];
  RELEASE(classManager);
  RELEASE(classEditor);
  RELEASE(hidden);
  RELEASE(window);
  RELEASE(filesOwner);
  RELEASE(firstResponder);
  RELEASE(fontManager);
  NSFreeMapTable(objToName);
  RELEASE(documentPath);
  RELEASE(savedEditors);

  RELEASE(selectionBox); 
  RELEASE(scrollView);
  RELEASE(classesScrollView);
  [super dealloc];
}

We should not call RELEASE(window) since it'll actually be released when 
calling performClose: (actually, in NSWindow: -close called by NSWindow: 
-performClose:).

Re: Include *Info.plist in bundle.make

I added this. Thanks.

Stefan Urbanek wrote:

> Hi,
> 
> In bundle.make there is around line 251:
> 
> # GNUstep bundles
> $(BUNDLE_DIR_NAME)/Resources/Info-gnustep.plist: $(BUNDLE_DIR_NAME)/Resources
> 	 <at> (echo "{"; echo '  NOTE = "Automatically generated, do not edit!";'; \
> 	  echo "  NSExecutable = \"$(INTERNAL_bundle_NAME)${BUNDLE_OBJ_EXT}\";"; \
> 	  if [ "$(MAIN_MODEL_FILE)" = "" ]; then \
> 	    echo "  NSMainNibFile = \"\";"; \
> 	  else \
> 	    echo "  NSMainNibFile = \"$(subst .gmodel,,$(subst .gorm,,$(subst
> .nib,,$(MAIN_MODEL_FILE))))\";"; \
> 	  fi; \
> 	  echo "  NSPrincipalClass = \"$(PRINCIPAL_CLASS)\";"; \
> 	  echo "}") >$ <at> 
> 
> 
> This has to be added to include *Info.plist (same as in application.make):
> 
> 	   <at> if [ -r "$(INTERNAL_bundle_NAME)Info.plist" ]; then \
> 	    plmerge $ <at>  $(INTERNAL_bundle_NAME)Info.plist; \
> 	  fi
> 

--

-- 

~/GNUstep/.GNUstepDefaults question

Hi,

Is there any reason why we don't enforce the file access permission 0600 
on ~/GNUstep/.GNUstepDefaults?

I think we should since most GNUstep apps might store sensitive 
informations in this file.

If we want to 'fix this', I have a patch that can be used.

Thanks,
		Ludovic

--

-- 
Live as if you were to die tomorrow.
Learn as if you were to live forever.
 - Gandhi

patch to common.make for cygwin paths

Hi,

There was a bug left in make/common.make when using Windows
paths (c:/GNUstep/...) because common.make tried to check
that it could find $(GNUSTEP_SYSTEM_ROOT) in $PATH. The last
set of Windows patches change this behaviour slightly:
GNUSTEP_SYSTEM_ROOT is allowed to have c:/... style paths,
but the paths get mangled when they are added to PATH so
that the colons don't break the path field separators.

Therefore the check in common.make needs to reflect this.

I've put the patch through a couple of tests on Windows: the
"cygpath" shell command is ONLY called when the path matches
^[a-zA-Z]:

Cheers,
Stephen Brandon
stephen <at> brandonitconsulting.co.uk

Index: make/common.make
===================================================================
RCS file: /cvsroot/gnustep/gnustep/core/make/common.make,v
retrieving revision 1.82
diff -r1.82 common.make
65c65,72
<   ifeq ($(findstring $(GNUSTEP_SYSTEM_ROOT)/Tools,$(PATH)),)
---
>   IS_WIN_PATH = $(shell echo "$(GNUSTEP_SYSTEM_ROOT)" | sed 's/^\([a-zA-Z]:.*\)//')

use of floorf rather than floor in NSTableView.m

In gui/Source/NSTableView.m there is use of the function
"floorf". I think it's the only time it's used in gnustep.
It is not found on Windows 2000 with CygWin/MinGW.

Is there any reason why it can't just be "floor"?

If not, could I please log this as a bug and request the
change:

diff -r1.44 NSTableView.m
2270c2270
<                 int remainingInt = floorf(remainingWidth);
---
>                 int remainingInt = floor(remainingWidth);

Cheers,
Stephen Brandon
stephen <at> brandonitconsulting.co.uk

Re: use of floorf rather than floor in NSTableView.m

> In gui/Source/NSTableView.m there is use of the function
> "floorf". I think it's the only time it's used in gnustep.
> It is not found on Windows 2000 with CygWin/MinGW.
> 
> Is there any reason why it can't just be "floor"?

I don't think there is any one. I definetly should have written
"floorf" when I wrote this line. 

> 
> If not, could I please log this as a bug and request the
> change:
> 
> diff -r1.44 NSTableView.m
> 2270c2270
> <                 int remainingInt = floorf(remainingWidth);
> ---
> >                 int remainingInt = floor(remainingWidth);

I'll commit this change ASAP.

bye,

Pierre-Yves Rivaille

Re: use of floorf rather than floor in NSTableView.m

Please ignore my previous mail, I was a little bit confused.

> In gui/Source/NSTableView.m there is use of the function
> "floorf". I think it's the only time it's used in gnustep.
> It is not found on Windows 2000 with CygWin/MinGW.
> 
> Is there any reason why it can't just be "floor"?

Here are the different floor functions from the linux manpage
       double floor(double x);
       float floorf(float x);
       long double floorl(long double x);

As the argument we pass to "floor" is a float, it seems logical to use
the "floorf" functions rather than the "floor" function.
But
1. floor is POSIX, floorf & floorl are from C99 (i'm just reading the
    manpage) (I don't really know what we can conclude from this)
2. using "floor" instead of "floorf" does produce the expected result
    (there is maybe a very small performance cost due to the conversion
     from float to double)

> If not, could I please log this as a bug and request the
> change:
> 
> diff -r1.44 NSTableView.m
> 2270c2270
> <                 int remainingInt = floorf(remainingWidth);
> ---
> >                 int remainingInt = floor(remainingWidth);