Re: Definition of Semantic: was What Lua can do that other programming languages can't do?

[Tim Hill <drtimhill@...>]

I would say a "semantic" bug is any deviation from expected behavior of a program that is grammatically correct (i.e. has correct syntax according to the formal grammar of the language). This of course leaves huge wiggle room for what "expected behavior" might mean, but I don't see any reason around this since, from a computers standpoint, semantic bugs cannot exist.

--Tim


On Mar 25, 2013, at 12:53 PM, Steve Litt <slitt@troubleshooters.com> wrote:

> On Fri, 8 Mar 2013 11:48:05 -0500
> Steve Litt <slitt@troubleshooters.com> wrote:
> 
>> On Fri, 08 Mar 2013 10:56:06 -0200
>> Jorge <xxopxe@gmail.com> wrote:
>> 
>>> On 03/07/2013 07:36 PM, Steve Litt wrote:
>>>> Speaking of tarpits, what the heck does "semantic" mean?
>>> 
>>> If I was a computer, mi mind would have had a stack overflow right
>>> there.
>> 
>> LOL.
>> 
>> Also, if you were a computer, you'd reject all the silly uses of the
>> word that seem to pop up every time someone talks code and tries to
>> sound smart.
>> 
>> Perhaps this is the solution to the stack overflow:
>> 
>> 
>> function semantic_usage_correct(usage)
>> 	if makes_sense(usage) == "definitely"
>> 		return(0)
>> 	elseif makes_sense(usage) == "no way"
>> 		return(1)
>> 	elseif you_are_tired_of_this_krap(usage)
>> 		return(-1)
>> 	else
>> 		usage = get_more_info(usage)
>> 		return(semantic_usage_correct(usage))
>> 	end
>> end
>> 
>> The thing that keeps you from stack overflowing is the test for
>> you_are_tired_of_this_krap(usage) :-)
>> 
>> Now, could somebody please code the makes_sense() function?
> 
> Hi all,
> 
> Watching this presentation:
> 
> https://www.usenix.org/conference/fast13/study-linux-file-system-evolution
> 
> Time marks 9:10 to 9:58, it looks to me like he's defining "semantic
> bugs" as incorrect coding of algorithms or data structures, or, at a
> higher thought level, failure to match the code to the problem
> domain.
> 
> Of course, that's in the tool's source code, which isn't the typical
> way I hear the words "semantic" and "semantics" used by developers.
> 
> Perhaps developers are using the word "semantics" to mean "behavior".
> For instance, the *syntax* of Lua tables is:
> 
> my_table={}                     -- Create empty table
> my_table.fname = "Steve"        -- Give value to "fname" key
> my_table["lname"] = "Litt"      -- Same behavior as above, different
> syntax table.insert(my_table,"one")    -- Set key 1 
>                                       -- (first numeric key)
>                                       -- to "one" 
> table.insert(my_table,"two")    -- Set next numeric key to "two"
> 
> The *behavior* (semantics?) of Lua tables are:
> 
> * Tables are key/value pair lists
> * Keys can be almost anything
> * Values can be anything
> * Tables passed and referenced by reference, not value
> * Tables whose only keys are 1 through a larger number can be iterated
>  a special way and treated like arrays in other languages
> 
> Have I made this distinction correctly? Is "semantics of" a synonym for
> "behavior of"?
> 
> Thanks,
> 
> SteveT
> 
> Steve Litt                *  http://www.troubleshooters.com/
> Troubleshooting Training  *  Human Performance
>