238501 - GetTextCharset is unreliable for bitmap fonts on NT3.51

Status: RESOLVED WONTFIX

(Core Graveyard :: GFX: Win32, defect)

Description

[[e-mail]]

I've found an ANSI font for which GetTextCharset returns OEM_CHARSET. Since
GetTextMetrics returns a .tmCharSet of ANSI_CHARSET for the same font, I suggest
that we switch to using GetTextMetrics instead.

Comment 1

[[e-mail]]

Attachment: Proposed patch

Comment 2

[timeless]

Comment on attachment 144656 [details] [diff] [review]
Proposed patch

>+    GetTextMetrics(aDC, &tm);
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/gdi/fontext_6r
lf.asp
Return Values

If the function succeeds, the return value is nonzero.

If the function fails, the return value is zero. 

Please check the return value (i know lots of our code doesn't, that's not a
good excuse)

Comment 3

[[e-mail]]

Actually, that's somewhat unlikely... the function called another function that
passed a GetTextMetrics call shortly before arriving here. I can investigate
folding the two GetTextMetrics calls together if you like.

Comment 4

[Christian :Biesinger (don't email me, ping me on IRC)]

Comment on attachment 144656 [details] [diff] [review]
Proposed patch

You don't need the & (~0xFF)) check anymore? 

thanks for making this patch.

Comment 5

[[e-mail]]

(In reply to comment #4)
>(From update of attachment 144656 [details] [diff] [review])
>You don't need the & (~0xFF)) check anymore? 
No, metrics.tmCharSet is a BYTE.

Comment 6

[[e-mail]]

Attachment: Alternative

This takes the call to GetTextMetrics out of GetNAME and into the caller...

Comment 7

[rbs]

I cannot vouch for this patch based on a single example, and no other history on
the matter.

Why, then, should GetTextMetrics() be more trustworthy than GetTextCharset() in
general? Because of that single font? What if there is another font which causes
GetTextMetrics() to be wrong and GetTextCharset() to be correct. There are cases
of badly designed fonts, and your font could be one. Since there are tons of
legacy code out there that depend on GetTextCharset(), solid evidence/history
(e.g., a KB article) is needed to consider it unreliable.

BTW, your alternative is not equivalent. Suppose that the new GetNAME() fails
_latter_, for example on the line with buffer.EnsureElemCapacity(), the caller
will not have the information that it is "other error" anymore.

Comment 8

[[e-mail]]

Attachment: Font in questions

(In reply to comment #7)
>There are cases of badly designed fonts, and your font could be one.
I'd better complain to Microsoft for badly designing their font then?

(From update of attachment 144744 [details] [diff] [review])
The new GetNAME is only called for truetype fonts so it does not have to
distinguish the non-truetype "error" from the other error cases.

Comment 9

[Ere Maijala (slow)]

Comment on attachment 144656 [details] [diff] [review]
Proposed patch

I see no problem in using GetTextMetrics. The font is apparently ok as it works
fine in WinXP, but NT 3.51 has a problem with GetTextCharSet. r=ere

Comment 10

[[e-mail]]

I was confused, because my WordPad does not list the font as a bitmap font.
Trust Microsoft to make my life difficult ;-)

Comment 11

[rbs]

Comment on attachment 144656 [details] [diff] [review]
Proposed patch

Needs more info.

Comment 12

[[e-mail]]

Attachment: Alternative updated for bitrot

Comment 13

[Gordon P. Hemsley [:GPHemsley]]

This bug has been buried in the graveyard and has not been updated in over 5 years. It is probably safe to assume that it will never be fixed, so resolving as WONTFIX.

[Mass-change filter: graveyard-wontfix-2014-09-24]