[pet] Poll to identify actively used functionality in PET

Xuchen Yao xuchen at coli.uni-saarland.de
Tue Jul 6 21:27:57 CEST 2010


May I add a very small feature request that requires code change of only 
one line?

I'm now running PET in "server" mode by sending input to PET's stdin and 
capturing output from PET's stderr. But the problem is that my code 
can't tell when it's the end of a parsing result from PET's stderr. 
Currently I modified the interactive() function from the CM branch of 
cheap.cpp by adding something at the end of a whole bunch of message, 
such as:

fprintf(fstatus, "cheap is brutally patched\n");
fflush(fstatus);

Of course this isn't good at all. So is it possible to add a dumb 
control character instead? I suggest 0x25:

0x25 End of medium         Intended as means of indicating on paper or 
magnetic tapes that the end of the usable portion of the tape had been 
reached.

http://en.wikipedia.org/wiki/C0_and_C1_control_codes

It's not used any more so won't break any other programs as 
Ctrl-C/Ctrl-D do.

If this is supported, then any Java/Python program can run PET in a 
pseudo "server" mode. Thanks.

Xuchen

On 2010-7-6 19:17, Bernd Kiefer wrote:
> This mail adresses all people using PET.
>
> We (the developers of PET) are planning major code changes that include
> also the removal of unused or unsupported functionality. We ask for
> your kind contribution to get a better picture of which of it is still
> in use and maybe needs to be supported. If you still actively use one
> of the following, please let us know.
>
> Input formats we'd like to discard:
>
> - pic / pic_counts
> - yy_counts
> - smaf
> - fsr
>
> Diverse options and the functionality behind it (some of them maybe more
> controversial):
>
> -default-les=traditional  determine default les by posmapping for all
>                           lexical gaps
> -no-chart-man             Old style chart manipulation aka chart
>                           dependencies
> -no-hyper                 Disable hyper-active parsing
> -no-filter                Disable automatically computed rule filter.
> -key=n                    Selecting the method for determining the rule
>                           argument that will be filled first
> -lattice (allows combination restrictions specifying path numbers)
> -server (yy -- Stephan?)
>
> The removal of the following functionality has already been agreed
> upon by a lot of people, still, you may also give feedback to that:
>
> - MRS modes that are only supported using Lisp code from the LKB with
>   ECL Modes that will still be supported are: native MRX and simple MRS
>   Conversion to rmrs formats is planned to be supported by a 
> standalone binary
> - FSPP input mode (requires ECL)
>
> Because these are the only modules that require the inclusion of ECL,
> support for ECL in PET will also be removed.
>
> Please take the time for answering us because we don't want to
> create unnecessary problems by removing important functionality.
>
> Best,
>     Bernd Kiefer
>



More information about the pet mailing list