Re: user: fix DrawTextExA/W on empty strings.

[Juan Lang]

Hi Rein,

-    if (!str) return 0;
+    if (!str || count == 0) return 0;
     if (count == -1) count = strlenW(str);
-    if (count == 0) return 0;
+    if (count == 0) {
+        if( flags & DT_CALCRECT) {
+            rect->right = rect->left;
+            rect->bottom = rect->top;
+        }
+        return 0;
+    }

This block will never be reached because of the if (!str || count == 0)
test above.

--Juan

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com

user: fix DrawTextExA/W on empty strings.

[Rein Klazes]

On Wed, 25 Jan 2006 07:49:57 -0800 (PST), you wrote:
> Hi Rein,
> - if (!str) return 0;
> - if (!str || count == 0) return 0;
> if (count == -1) count = strlenW(str);
> - if (count == 0) return 0;
> - if (count == 0) {
> -
>    if( flags & DT_CALCRECT) {
> -
>        rect->right = rect->left;
> -
>        rect->bottom = rect->top;
> -
>    }
> -
>    return 0;
> - }
>
> This block will never be reached because of the if (!str || count == 0)
> test above.

Wrong, look again.

It can be coded a little bit more efficiently though. I will make a
different patch.

Rein.

Rein.

user: fix DrawTextExA/W on empty strings.

[Dimi Paun]

From: "Rein Klazes"
> Wrong, look again.

You must be kidding. Care to explain how 'stmt'
can ever be executed in:

...
   if (A || B) return;
   if (B) stmt;
...

user: fix DrawTextExA/W on empty strings.

[Paul Vriens]

On Wed, 2006-01-25 at 14:03 -0500, Dimi Paun wrote:
> From: "Rein Klazes"
> > Wrong, look again.
>
> You must be kidding. Care to explain how 'stmt'
> can ever be executed in:
>
> ...
>    if (A || B) return;
>    if (B) stmt;
> ...

But there's an if in between:

if (count == -1) count = strlenW(str);

Paul.

user: fix DrawTextExA/W on empty strings.

[Dimi Paun]

From: "Paul Vriens"
> But there's an if in between:
>
> if (count == -1) count = strlenW(str);

Yeap, missed that one. Which proves that the code
should be improved, since it managed to confuse 2
people already ;)

user: fix DrawTextExA/W on empty strings.

[Bill Medland]

On January 25, 2006 10:49 am, Rein Klazes wrote:
> On Wed, 25 Jan 2006 07:49:57 -0800 (PST), you wrote:
> > Hi Rein,
> > - if (!str) return 0;
> > - if (!str || count == 0) return 0;
> > if (count == -1) count = strlenW(str);
> > - if (count == 0) return 0;
> > - if (count == 0) {
> > -
> >    if( flags & DT_CALCRECT) {
> > -
> >        rect->right = rect->left;
> > -
> >        rect->bottom = rect->top;
> > -
> >    }
> > -
> >    return 0;
> > - }
> >
> > This block will never be reached because of the if (!str ||
> > count == 0) test above.
>
> Wrong, look again.
>
> It can be coded a little bit more efficiently though. I will
> make a different patch.
>
> Rein.
>
> Rein.

(Apart from the other discussion) are you sure that you really 
mean that?
"If the length is explicitly stated to be zero then don't do the 
calcrect but if the length is to be measured and happens to come 
out as zero then do do the calcrect"
If so then I suggest at least a comment that it is deliberate; it 
sounds counterintuitive to me.

user: fix DrawTextExA/W on empty strings.

[Rein Klazes]

On Fri, 27 Jan 2006 10:28:04 -0800, you wrote:
> (Apart from the other discussion) are you sure that you really 
> mean that?
> "If the length is explicitly stated to be zero then don't do the 
> calcrect but if the length is to be measured and happens to come 
> out as zero then do do the calcrect"
> If so then I suggest at least a comment that it is deliberate; it 
> sounds counterintuitive to me.

I thought I did even better by providing tests for each of those cases,
which pass here under Win2k and XP.

Other weird observations: the different behavior of the A and W
functions; the "return 0" in that code are actually "return 1" on Win2k
but not on XP.

Rein.