[dpdk-dev] [PATCH] Add external parser support for unknown commands.

[Neil Horman]

On Tue, Nov 04, 2014 at 02:44:39PM +0000, Wiles, Roger Keith wrote:
> 
> > On Nov 4, 2014, at 5:27 AM, Neil Horman <nhorman at tuxdriver.com> wrote:
> > 
> > On Tue, Nov 04, 2014 at 04:52:48AM +0000, Wiles, Roger Keith wrote:
> >> 
> >>> On Nov 3, 2014, at 5:42 PM, Neil Horman <nhorman at tuxdriver.com> wrote:
> >>> 
> >>> On Mon, Nov 03, 2014 at 03:26:50PM -0800, Stephen Hemminger wrote:
> >>>> On Mon, 3 Nov 2014 16:50:15 +0000
> >>>> "Wiles, Roger Keith" <keith.wiles at windriver.com> wrote:
> >>>> 
> >>>>> 
> >>>>>> On Nov 3, 2014, at 10:06 AM, Neil Horman <nhorman at tuxdriver.com> wrote:
> >>>>>> 
> >>>>>> On Mon, Nov 03, 2014 at 02:25:51PM +0000, Wiles, Roger Keith wrote:
> >>>>>>> 
> >>>>>>>> On Nov 3, 2014, at 8:16 AM, Bruce Richardson <bruce.richardson at intel.com> wrote:
> >>>>>>>> 
> >>>>>>>> On Mon, Nov 03, 2014 at 02:08:46PM +0000, Wiles, Roger Keith wrote:
> >>>>>>>>> 
> >>>>>>>>>> On Nov 3, 2014, at 4:41 AM, Bruce Richardson <bruce.richardson at intel.com> wrote:
> >>>>>>>>>> 
> >>>>>>>>>> On Sun, Nov 02, 2014 at 04:28:28PM -0600, Keith Wiles wrote:
> >>>>>>>>>>> Allow for a external parser to handle the command line if the
> >>>>>>>>>>> command is not found and the developer has called the routine
> >>>>>>>>>>> int cmdline_set_external_parser(struct cmdline * cl,
> >>>>>>>>>>>                            cmdline_external_parser_t parser);
> >>>>>>>>>>> function to set the function pointer.
> >>>>>>>>>>> 
> >>>>>>>>>>> The function for the external parser function should return CMDLINE_PARSE_NOMATCH
> >>>>>>>>>>> if not able to match the command requested or zero is handled.
> >>>>>>>>>>> 
> >>>>>>>>>>> Prototype of external routine:
> >>>>>>>>>>> int (*cmdline_external_parser_t)(struct cmdline * cl, const char * buy);
> >>>>>>>>>>> 
> >>>>>>>>>>> Signed-off-by: Keith Wiles <keith.wiles at windriver.com>
> >>>>>>>>>> 
> >>>>>>>>>> Hi Keith,
> >>>>>>>>>> 
> >>>>>>>>>> what is the expected use case for this? Is it for embedding other programming languages alongside the existing DPDK command-line or some other purpose? [Perhaps the use case could be called out in the patch description]
> >>>>>>>>> 
> >>>>>>>>> Hi Bruce,
> >>>>>>>>> 
> >>>>>>>>> I guess the external parser could be used for other programming languages, but the case I was looking at was to provide a default escape from the command line parser to allow my application to handle the commands not understood by the parser. Now that you point it out I could use something like ‘%<line-of-script-code>’ to execute a single line of script code, which is a good idea (thanks).
> >>>>>>>>> 
> >>>>>>>>> One case I am looking at is when you want to execute a command and do not want to add the support into the commands.c file for every possible command. Take the case where you have a bunch of scripts (Lua) in a directory much like a bin directory. Then you could type foo.lua or foo on the command line and execute the foo.lua having the application detect you want to load and run a Lua script after it has finished parsing for the builtin commands.
> >>>>>>>>> 
> >>>>>>>>> For Pktgen I had to add a command called ‘run <filename> <args…>’ to support running a script with arguments. I also needed to add a argvlist type to cmdline to not error out on that command and split up the args into a argv list like format. (Maybe I need to submit that code??) It seemed more straight forward to just pass the command line to the application to run the command. I understand that seems like a minor point, but it does make it easier to use and to support the features I want to support in my PoC.
> >>>>>>>>> 
> >>>>>>>>> Using this method you can just type the name instead of something like ‘run foo.lua’ or just ‘run foo’ and let the code figure out what to run. I have more plans for this features as well and have not finished the basic PoC yet. If you want a peek I can show you what I am working on currently.
> >>>>>>>>> 
> >>>>>>>>> Does this help and do I really need to add all of this to the commit message :-)
> >>>>>>>>> 
> >>>>>>>> Thanks for the explanation. However, if you are looking to have the application handle a bunch of commands itself, why does it need to use the commandline library at all? Why not just have the app handle all the commands instead of some of them?
> >>>>>>> 
> >>>>>>> I guess that would be reasonable, but then I would have to add support for all of the command line parsing being done in the cmdline code. Think of this as a default case for the parser and to me that makes more sense then just doing my own command line design. In the cmdline code you guys provided is a lot of features like history, control key support, arg parsing (IP, MAC) and many others. I would rather not have to write that code myself.
> >>>>>>> 
> >>>>>>> The default case is the same behavior today, with giving a no match error unless they add the external parser.
> >>>>>> 
> >>>>>> It seems alot simpler than that to me.  Looking at the test applications, the
> >>>>>> command line parser expects the application to create an array of
> >>>>>> cmdline_parse_ctx_t structures to support new option parsing.  If your goal is
> >>>>>> to support other languages, it seems to make more sense to just use foreign
> >>>>>> language bindings to merge your coding language support with the DPDK
> >>>>>> (ostensibly you will already have to do that if you want to use other parts of
> >>>>>> the DPDK).
> >>>>> Hi Neil,
> >>>>> 
> >>>>> A true language binding like Lua or one of those other languages :-) you are correct to believe binding directly using ‘C’ code is the right solution . In Pktgen I use Lua as the direct language binding and extend Lua with specific Pktgen functions.
> >>>>> 
> >>>>> What I am doing here is to add a default case to cmdline code, which just happens to allow me to parse the cmdline in the application. Being able to execute say a line of script code is not really the requirement IMO. Being able to extend the cmdline code with a default case is a good feature and allows the developer to extend cmdline for some simple cases. The cmdline code is kind of simple, but does require a fair amount of structures, code and understanding to write a complex extendable command line interface. It does seem hard to find a clean, simple and usable embedded command line code base is not very easy to locate. 
> >>>>> 
> >>>>> Adding a true language binding really requires using code to extend the language as I did with Lua and Pktgen. It could have been done with any language I just picked Lua, but the patch does not really add support for a language other then giving some support for someone to handle the no_match case.
> >>>>> 
> >>>>> The use case for this feature is not just for Pktgen, but another solution I hope everyone will find useful when I get it more complete.
> >>>>> 
> >>>>> Thanks
> >>>>> ++Keith
> >>>>> 
> >>>>> PS. on a different topic I was thinking about suggesting and writing a patch to add Lua with DPDK specific binding and extensions. (also allowing those `other` languages too :-) Being able to use a scripting language and be able to call DPDK API’s could be useful. How useful not sure at this time. (If you want to talk about this topic please start a new thread).
> >>>>>> 
> >>>>>> Am I missing something?
> >>>>>> Neil
> >>>>>> 
> >>>>>> 
> >>>>>>>> 
> >>>>>>>> /Bruce
> >>>>>>> 
> >>>>>>> Keith Wiles, Principal Technologist with CTO office, Wind River mobile 972-213-5533
> >>>>> 
> >>>>> Keith Wiles, Principal Technologist with CTO office, Wind River mobile 972-213-5533
> >>>> 
> >>>> I wouldn't invest a lot of sweat in the command line parser.
> >>>> The one in the DPDK is "good enough" for what it needs to do, but really isn't
> >>>> very complete and flexible. Seems like the kind of thing that doesn't really even
> >>>> need to be in DPDK. Better off being part of some other library.
> >>>> 
> >>> Well, something needs to be there to parse the libraries' common options, though
> >>> I agree, making eal_cmdline just a registration frontend to getopt or
> >>> getopt_long would be sufficient.
> >> 
> >> Until we have a better command line solution, which I think would be great, but in the mean time I would like to see this patch applied if no one has a technical reason or better suggestion.
> >> 
> >> I think this patch is fairly simple and I think we need a way to handle the default case. If someone could please review the patch, that would be great.
> >> 
> > I have an objection, specifically, that its not necessecary.  You can already
> > accomplish what you want to do by adding structures to the context array in the
> > cmdline structure.  I realize its not as easy as just adding an external parser
> > function, but its the designed way to add options.  This does little more than
> > add addition API surface without any real need.
> > Neil
> 
> Neil
> 
> I do not agree with your comments as I see it to be a small extension to cmdline to handle the case where I will have to possibly add a huge number of commands/code to the cmdline structures. Using this method I am able to add these very simple commands without having to add more code for this use case.
> 
Can you provide a real example here?  usnig vague terms like "huge" really makes
more of an emotional argument than a factual one.  To cite an example the
cmdline_test program adds a command line paramter that just lets you parse a
number out of the command line.  Silly, granted, but it serves the purpose.  Its
called cmd_num, and with functions and data all told, it looks like it takes
about 17 lines of code.  Now thats more than what you're adding with you patch,
I grant you, but I assert that the "potentially huge" argument you're making
above is false, especially whan you consider that some reasonably clever coding
can likely allow you to reuse function parsing fairly easily.

> Lets say you have a directory on the disk that has possibly a 100 little commands, without this minor change I would have to write 100 little structures/code for cmdline to handle each case. Another option is to write a single command to handle these commands. I used this method in Pktgen and could do ‘run foo <args>’ style commands, but it would be much simpler for the user to just type ‘foo <args>’ instead.
> 
Yes, but thats honestly true of every command line parser, something needs to
define the tokens that identify the option, the function to handle its
interpretation and the structures to glue it all together.  That can be in the
DPDK, or in something else.  To use your example above, one presumes that your
100 little commands all have some sort of parsing structure coded elsewhere,
along with the code to do that parsing, right? You don't seem to take that
additional code into account here.  If you can remove that parsing code from
your application, then it seems to me the additional 17 lines above really isn't
that big a deal.

It seems to me that, what this boils down to is the fact that you have an
application that needs to parse a single command line with two different parser,
whcih is just kind of a lousy situation, because parsers all assume that they
are the only thing parsing the command line.  Honestly, I really question the
need for a command line parser so tightly integrated with DPDK at all.
rte_eal_init really shouldn't be acepting a straight command line buffer.  As a
library, it should more likely accept a configuration structure which it just
doles out to individual components during initalization.  That can leave
applications like yours to handle command line parsing 100% as you see fit, and
then build the DPDK configuration from that as you like.

> Having a default handler for commands just makes a lot of sense to me and I do not buy the 'added API surface without any real need' statement.
I'm not sure whats not to buy there.  I see the above as facts:

1) You're adding API surface (you added a function that is exported from the
library)

2) You don't need to do (1) (theres an already existing alternate method to do
what you want)

You can argue all day that your method is better, but it doesn't change the fact
that (1) and (2) are true. And I don't want to go adding additional methods to
the existing ones as there will be a need to support them in the future, and as
such, If we're going to start deprecating API's in favor of superior designs,
I'd like to do that as infrequently as possible.
Neil
 
> 
> Thanks
> ++Keith
> > 
> > 
> >> ++Keith 
> >> 
> >>> 
> >>> Neil
> >>> 
> >> 
> >> Keith Wiles, Principal Technologist with CTO office, Wind River mobile 972-213-5533
> 
> Keith Wiles, Principal Technologist with CTO office, Wind River mobile 972-213-5533
>