[PATCH:libx11] parse_omit_name should increase num of fields before appending encoding

[Hong Xu]

Hong Xu <hong at topbug.net> writes:

> Alan Coopersmith <alan.coopersmith at oracle.com> writes:
>
>> On 03/15/15 11:39 PM, Hong Xu wrote:
>>> parse_omit_name currently does not increase number of fields to 12 after
>>> appending encoding, This does not lead to the correct font is a xlfd
>>> like "-*-*-medium-r-*-*-33-*" is given.
>>>
>>> Signed-off-by: Hong Xu <hong at topbug.net>
>>>
>>> ---
>>> Please note that the "if" in this patch is replaced by "while". The
>>> reason is that I found the commit 3d69b0a83e62f8f6fbd changed the
>>> "while" to "if", which does not seems logically correct for me and
>>> actually accidently "fixed" the bug
>>> http://sourceware.org/bugzilla/show_bug.cgi?id=10948
>>
>> Multiple people found and fixed that bug - I am not going to revert
>> the change and destroy performance without a better reason than one
>> person not understanding the fix.
>>
>> It's not just the bug you cite, but was also a bug our Solaris i18n team
>> found and fixed:
>>     https://hg.java.net/hg/solaris-x11~x-s12-clone/rev/6fe48a3a12bf
>> and one Thomas Dickey found and fixed while investigating xterm performance:
>>     http://invisible-island.net/xterm/xterm.faq.html#slow_menus
>
> From what you said, I think I can conclude 2 problems:
>
>    1. -*-*-medium-r-*-*-33-* cannot be correctly parsed;
>    2. performance.
>
> This patch fixes problem 1 (for me); for problem 2, this patch does not
> simply revert the change -- it also removed the line
>
>     strcpy(last + 2, font_data->name);
>
> in the loop, which is possibly the performance hoop (because I cannot
> reproduce any performance issue, I can only guess).
>
> It would be best if you can test whether the performance issue is fixed
> ...

I've sent a second version of this patch, which also removes the
get_font_name line in the loop. The approach also changed a bit: all
missing fields are filled by *- first as well as the final
font_data->name, then call get_font_name, rather than append
font_data->name and call get_font_name after appending each *-. This
should have a much better performance.

Hong

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://lists.x.org/archives/xorg-devel/attachments/20150318/da3bfcd9/attachment.sig>