OSSP CVS Repository

ossp - ossp-pkg/xds/TODO 1.16
Not logged in
[Honeypot]  [Browse]  [Directory]  [Home]  [Login
[Reports]  [Search]  [Ticket]  [Timeline
  [Raw

ossp-pkg/xds/TODO 1.16
TODO LIST AND MILESTONES
========================

o M1 (peti) [11-Jun-2001]
 - evaluation checklist for XDS and AMP
   done: 100% [-> EVAL]

o M2 (peti+rse) [25-Jun-2001]
 - design/architecture paper (authentication+encryption framework)
   (< 10KB Text + 1x Figure)
   done: 30% [-> architecture.fig]

o M3 (peti+rse) [2-Jul-2001]
 - design libxds API
   done: 100% [-> libxds/xds.h + libxds/xds.pod]

o M4 (peti) [4 weeks until 30-Jul-2001]
 - implementation of libxds library framework
   done: 100% [-> libxds/*.c]

 - implementation XDR engines for basic data types.
   done: 99% [-> everything is implemented except "double" support]

 - implementation XML engines for basic data types.
   done: 99% [-> everything is implemented except "double" support]

 - regression tests for libxds (make check)
   done: 100% [-> regression-tests/*.c]

 - 16bit Integers mean trouble when passing them through "...":

     | xdr-encode-uint16.c:46: `xds_uint16_t' is promoted to `int' when passed through `...'
     | xdr-encode-uint16.c:46: (so you should pass `int' not `xds_uint16_t' to `va_arg')

   Maybe we should skip int16 support at all? Currently I work around
   this problem by implicitely assuming that the short is promoted to
   an int.

   To make matters even worse: The RFC doesn't even define how XDR
   deals with 16-bit integers! The mere existance of such a datatype
   is not positively confirmed.

  - XML: <int32>1234</int32>
         uint32, int64, uint64,

       - xml_encode_begin():
	 <?xml version="1.0" encoding="UTF-8" standalone="no">
         <!DOCTYPE xds SYSTEM "http://www.ossp.org/pkg/xds/xds-xml.dtd">
         <xds>

       - xml_encode_end():
         </xds>
 - UFT-8 RFC: rfc2279.txt

    Escape: <   <--> &lt;
            &   <--> &amp;
            ]]> <--> ]]&gt;

 - strcpy() copies one byte more than the length of the text itself
   because the trailing \0 is written to. Currently, this behavior is
   not taken into account when calculating the correct buffer size in
   the xml-encode-*-.c modules. This has to be fixed!

  - libm may be used. We don't depend on sprintf() but convert out
    numbers ourselves for portability.

  - IEEE values ranges for floating point numbers will be used for
    accurarcy in XML presentation.

 - autoconf should find the types for int32 etc. even if they're not
   defined in sys/types.h.

 - autoconf should specify the compiler flags depending on what the
   compiler supports.

o M5 (peti+rse) [1 week until 06-Aug-2001]
  - [peti] documentation for libxds (xds.pod)
  - [rse]  OSSP adjustments for libxds

o M6 (peti+rse) [1 week until 13-Aug-2001]
  - design API for libamp

o M7 (peti) [4 weeks until 10-Sep-2001]
  - implementation of libamp library framework
  - implementation of libamp backends SHM, TCP and UDP
  - regression tests for libamp

o M8 (peti+rse) [1 week until 17-Sep-2001]
  - [peti] documentation for libamp
  - [rse]  OSSP adjustments for libamp


CVSTrac 2.0.1