Wireshark-dev: Re: [Wireshark-dev] Use val_to_str_ext() & friends to access sminmpec_values[]
Date Index · Thread Index · Other Months · All Mailing Lists
Date Prev · Date Next · Thread Prev · Thread Next
From: Bill Meier <wmeier@xxxxxxxxxxx>
Date: Thu, 23 Sep 2010 15:01:25 -0400
Bill Meier wrote:

Jakub Zawadzki wrote:

Hi,

On Fri, Sep 17, 2010 at 02:00:44PM -0400, Bill Meier wrote:

Anders Broman wrote:

Is the script "fixed" to fill out any "holes" in the list with "unasigned"?
That would be the most efficent use...
If I do this, match_strval() against the sminmpec list will no longer return NULL for a "missing" entry (other than "past the end"). 


Can we use NULLs instead of 'unasigned'?

All value_string_match_t functions (except of _match_strval_linear(), which is just wrapper to match_strval())
are using vs->length already.

It might not work when wireshark want to iterate through values.
I considered using NULL, but came to the conclusion that match_strval_ext doing a linear search could/would terminate early because of the NULL (which is what I think you are saying above).
However: I think you're on the right track: for "extended" value strings I'll just create a version of the linear match which uses vs->length instead of checking for NULL.
I decided to use "(Unknown)" instead of NULL for the string value for entries used to fill "gaps" in sminmpec.
The main reason: there's code that assumes that value_strings (even those referenced by value_string_ext) end with a NULL strptr.
I could have changed that code when dealing with value_string_ext, but decided I shouldn't change the (long standing) semantics of value_strings.
If the use of "(Unknown)" to fill in the gaps for sminmpec proves to be a problem for some reason, reverting will just cause a binary search to be used which isn't really a problem.