55450 - New Gfx implementation for Mozilla on BeOS

Status: VERIFIED FIXED

(Core :: Layout, defect, P3)

Description

[Yannick Koehler]

Based on the Gtk I've re-implemented the Gfx of the BeOS to properly render font and graphics again.  I'm including the diff with the option -wu because there's a lot of space stuff added so that it respect coding convention also.

All this stuff is beos specific, there's only two place where it's not and it's because I had to add the XP_BEOS keywords.

Comment 1

[Yannick Koehler]

Attachment: New Gfx for BeOS

Comment 2

[Yannick Koehler]

That patch need new file, so I'm including them as another attachment (zipped this time).

Comment 3

[Yannick Koehler]

Attachment: zip of the missing file (extract here mozilla/gfx/src/)

Comment 4

[buster]

Is the purpose of this bug to get your patch checked in?  If so, please add the
"patch" keyword, and get on the newsgroups to see who is willing to take this on
for you. Does anybody working on BeOS have CVS access?

Comment 5

[Yannick Koehler]

Yes, I'm on BeOS I have cvs access.  But the check-in rules state that this 
code should be reviewed by the module and a reviewers.  

Comment 6

[Yannick Koehler]

Another things, I just got my cvs account therefore I would like to get sure 
that I'm not doing things wrong.  Which is why I've not done the commit.  

Comment 7

[Yannick Koehler]

Attachment: another cvs diff -wu nsRenderingContext.cpp

Comment 8

[Yannick Koehler]

I've included another patch for nsRenderingContext.cpp to cover a change inside GetWidth()/DrawString() for utf8 string (we were counting the passed string length instead of the converted one.)

Comment 9

[cls]

With the exception of a single header, the changes are entirely within the beos
specific directories. r=cls

Comment 10

[Kevin McCluskey (gone)]

Dividing up claytons bugs to triage

Comment 11

[Christopher Blizzard (:blizzard)]

Can you use an nsCOMPtr for the mSurface and mOffscreenSurface?  There seems to
be a good bit of NS_ADDREF() and NS_RELEASE() going on here that could be
replaced with nice 'ol nsCOMPtrs.

Is using a GS pool going to get you anything on BeOS?  We use it on unix because
of the overhead of the server roundtrips but I thought that BeOS was local.

nsRenderingContextBeOS::SetLineStyle() doesn't look done:

 NS_IMETHODIMP nsRenderingContextBeOS::SetLineStyle(nsLineStyle aLineStyle)
 {
   if (aLineStyle != mCurrentLineStyle)
   {
-printf("nsRenderingContextBeOS::SetLineStyle not implemented!\n");
-#if 0
     switch(aLineStyle)
     {
       case nsLineStyle_kSolid:
-        ::gdk_gc_set_line_attributes(mSurface->gc,
-                          1, GDK_LINE_SOLID, (GdkCapStyle)0, (GdkJoinStyle)0);
+        {
+        }
         break;

The code violates a lot of our 80 column rules and is generally pretty messy in
terms of indentation style.  However, if it works better than what's in there
now I guess it's OK to check it in. r=blizzard

Comment 12

[Yannick Koehler]

I learned about nsCOMPtr and will actually use them.  But at the moment of that 
patch I didn't felt confident enough to apply that new knowledge.  As for the 
80 column stuff I know that, I'm working bit by bit to enhance the quality of 
the code layout and you'll see it progressing (actually I've worked hard on 
removing tabs and replacing it with 2 spaces as others codes)  Next I'll see 
what I can do with columns.

So I got two reviewers, good, I will sit down tonight and review again the 
check in procedure and then commit it if things are ok (green tree etc..)

Therefore I'm assigning this bug to myself

Comment 13

[Yannick Koehler]

I committed the patch tonight!  yeepee!

Comment 14

[cls]

Verified.