Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-23 Thread Joachim Trémouroux
Hi Michael,

Le 22 novembre 2010 22:07, Michael Meeks michael.me...@novell.com a écrit
:


Really, of course - I'd love to have someone working on eg.


 http://wiki.documentfoundation.org/Development/Easy_Hacks#don.27t_ship_150_duplicate_placeholder_icons

Which should be mind-numblingly simple and yet yield a real
 image-size
 (and hence performance) win :-)

Any chance of a small detour on the way ? :-)

Thanks anyhow,

Michael.

 --
  michael.me...@novell.com  , Pseudo Engineer, itinerant idiot




I will work on this. I see two possible ways:

- wrap current loadImage with something similar to this:
bool found = loadImage(...)
if (!found) {
  found = loadImage( default_icon.png ...)
}
return found

 - alternatively, add the default icon path to the list of paths that are
passed to the ImplImageTree::find method.
I think this will be a bit less efficient as the normal case is that the
icon exist. Your opinion?


Furthermore, the duplicate icons currently exist in several (4?) sizes. So
we could have 5 different icons:
lc_default_icon.png
lx_default_icon.png
sc_default_icon.png
sx_default_icon.png
default_icon.png
and based on the input name we can return an icon of the correct size.

Does it look ok for you?

Regards,
Joachim.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-23 Thread Pierre-André Jacquod
On 11/23/2010 09:23 AM, Joachim Trémouroux wrote:

 Hi Michael,
 I will work on this. I see two possible ways:
 

Ok Michael, then I will continue within binfilter... ::-))

But a additional question:

From Norbert Thiebaud:

 To be consistent, I rather see them commented out
 i.e

 foo(int bar)
 -
 foo(int /*bar*/)
 rather than
 foo(int)

To have a better code, would it not be better to change the prototyping
of the function from foo (int) to foo() ?? Or did I miss a point?

So would the code be cleaner or do you intend to drop the binfilter
components very soon ?
regards


___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-23 Thread Caolán McNamara
On Tue, 2010-11-23 at 21:05 +0100, Pierre-André Jacquod wrote:
 To have a better code, would it not be better to change the prototyping
 of the function from foo (int) to foo() ?? Or did I miss a point?

It would be better, *but* often these are virtual methods and have to
retain their signature to remain virtual, not always, but sometimes and
it can be tricky to get right. For the case of the binfilter its just
easier to guarantee lockin the current behaviour.

C.

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-22 Thread Michael Meeks
Hi Joachim,

On Mon, 2010-11-22 at 13:59 +0100, Joachim Trémouroux wrote:
 I have fixed some compilation warnings in binfilter and some
 unnecessary comments. Patch is attached for review.

Hokay - the binfilter is of course un-loved and under-maintained.

 I have two questions:
  - /*N*/ . Should I fix them massively? I have currently left them
 unchanged to avoid a big cleanup patch.

I guess this should be done by some large-scale sed at some point
centrally. Certainly not a useful patch for the list ;-)

  - Unused parameters in some methods are triggering compilation
 warnings. Should I fix them by removing the parameter name ?

Sounds reasonable.

Really though - the binfilter is not the best place to focus cleanups
(though I appreciate it is fugly old code ;-) Most people will not want
it (only useful for decade-old binary StarOffice file formats).

Thanks,

Michael.

-- 
 michael.me...@novell.com  , Pseudo Engineer, itinerant idiot


___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-22 Thread Norbert Thiebaud
2010/11/22 Michael Meeks michael.me...@novell.com:
 Hi Joachim,

 On Mon, 2010-11-22 at 13:59 +0100, Joachim Trémouroux wrote:
 I have fixed some compilation warnings in binfilter and some
 unnecessary comments. Patch is attached for review.

        Hokay - the binfilter is of course un-loved and under-maintained.

 I have two questions:
  - /*N*/ . Should I fix them massively? I have currently left them
 unchanged to avoid a big cleanup patch.

        I guess this should be done by some large-scale sed at some point
 centrally. Certainly not a useful patch for the list ;-)

  - Unused parameters in some methods are triggering compilation
 warnings. Should I fix them by removing the parameter name ?

        Sounds reasonable.

To be consistent, I rather see them commented out
i.e

foo(int bar)
-
foo(int /*bar*/)
rather than
foo(int)


        Really though - the binfilter is not the best place to focus cleanups
 (though I appreciate it is fugly old code ;-) Most people will not want
 it (only useful for decade-old binary StarOffice file formats).

as unloved as it maybe. as long as it _IS_ in the build, it has to build
and, especially in // build, having the build.log flooded with
thousands of warning kind of defeat the point of having warning in the
first place.

Maybe you could draw the outline of what need to be accomplish to make
binfilter a separate stand-alone project ?

Norbert


        Thanks,

                Michael.

 --
  michael.me...@novell.com  , Pseudo Engineer, itinerant idiot


 ___
 LibreOffice mailing list
 LibreOffice@lists.freedesktop.org
 http://lists.freedesktop.org/mailman/listinfo/libreoffice

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-22 Thread Pierre-André Jacquod
Hi,

  - Unused parameters in some methods are triggering compilation
 warnings. Should I fix them by removing the parameter name ?
 
   Sounds reasonable.
 
   Really though - the binfilter is not the best place to focus cleanups
 (though I appreciate it is fugly old code ;-) Most people will not want
 it (only useful for decade-old binary StarOffice file formats).

Okay, but at compile time, there is really a flow of warning. So I
wanted also to work a bit on it. That's just cleaning, not really
improving.

But when I see:

clone/filters/binfilter/bf_sc/source/core/tool/sc_token.cxx:1573:28:
warning: comparison is always false due to limited range of data type

and the source code is (it starts form line 1573..

   if( n  MAXSTRLEN-1 )
/*N*/  {
/*?*/  DBG_ERRORFILE( bad string array boundary );
/*?*/  USHORT nDiff = n - (MAXSTRLEN-1);
/*?*/ n = MAXSTRLEN-1;
/*?*/ rStream.Read( c, n );
/*?*/ rStream.SeekRel( nDiff );
/*N*/  }
/*N*/  else
/*N*/  rStream.Read( c, n );
/*N*/  cStr[ n ] = 0;

would as cleaning be acceptable to change the code to:

/*N*/  rStream.Read( c, n );
/*N*/  cStr[ n ] = 0;

??? I wonder...
If yes, a lot of code can be removed (between 40 and 60 comparisons are
generating this warning). So I would prefer not starting and beeing told
not to touch it afterward.

Thanks
Pierre-André

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: [Libreoffice] [patch] binfilter: cleanup and compilation warnings removal.

2010-11-22 Thread Michael Meeks
Hi Pierre,

On Mon, 2010-11-22 at 21:06 +0100, Pierre-André Jacquod wrote:
 Okay, but at compile time, there is really a flow of warning. So I
 wanted also to work a bit on it. That's just cleaning, not really
 improving.

Sure sure :-) fair enough.

 clone/filters/binfilter/bf_sc/source/core/tool/sc_token.cxx:1573:28:
 warning: comparison is always false due to limited range of data type

Right - so we can chop that code out with prejudice as you say :-)

 would as cleaning be acceptable to change the code to:
 
 /*N*/  rStream.Read( c, n );
 /*N*/  cStr[ n ] = 0;

Of course ! no point in keeping dead code around.

 If yes, a lot of code can be removed (between 40 and 60 comparisons are
 generating this warning). So I would prefer not starting and beeing told
 not to touch it afterward.

Go for it :-) of course, this is the sort of patch that scares patch
reviewers, so I would pair up the compile messages of this form, with
the diff (if you can) ?

Really, of course - I'd love to have someone working on eg. 

http://wiki.documentfoundation.org/Development/Easy_Hacks#don.27t_ship_150_duplicate_placeholder_icons

Which should be mind-numblingly simple and yet yield a real image-size
(and hence performance) win :-)

Any chance of a small detour on the way ? :-)

Thanks anyhow,

Michael.

-- 
 michael.me...@novell.com  , Pseudo Engineer, itinerant idiot


___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice