Pike-devel September 2007

pike-devel@lists.lysator.liu.se

5 participants
12 discussions

Pike conference 2007 - immediate release
by Matt Hardy - Webhaven via Martin Baehr 14 Sep '07

14 Sep '07

(forwarding this for matt, martin) Greetings, I am pleased to inform everyone concerned that the most preferred dates have been confirmed and facilities booked in Copenhagen, Denmark. 14 + 15 + 16 November 2007. First 2 days will be located at Copenhagen University. Last official day is at a carrier neutral datacentre also located in Copenhagen. Further days, for those who have time to stay longer, are TBA, depending on interest. I now hope everyone can look at calendars to ensure they have adequate time to attend the official program, it will be a very good chance to carry on discussions from last year in Riga, as well as discuss new developments undertaken since. All submissions for presentation or discussion topics are welcome. Please forward your submissions to the list, and/or you can attach documents by email to matt(a)webhaven.eu We will construct a 2007 website and publish a clear program for the 3 days. Pike Conference 2007 Copenhagen, Denmark ---->8------ if (Human.Programmer("meaning you")->wants_to_go()) { go(); } ----8<------

1 0

Structured data mk III
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 02 Sep '07

02 Sep '07

I've just made some experiments for a replacement to ADT.struct and ADT.Struct: > program struct = >> ADT.Serialize.SimpleStruct("struct", >> ADT.Serialize.Int32, "integer", >> ADT.Serialize.SimpleStruct("substruct", >> ADT.Serialize.String, "data" >> ), "sub"); > object o = struct(14); > indices(o); (1) Result: ({ /* 5 elements */ "encode", "decode", "_integer", "integer", "sub" }) > o->integer; (2) Result: 14 > o->sub; (3) Result: substruct() > o->sub->data = "foobar"; (4) Result: "foobar" A decent encoding API is easy: > o->encode(); (5) Result: "\0\0\0\16foobar" Now to the hard part; a streaming decoder API... As far as I know, there are two main approaches: * Pull -- the user sends a data source (eg a file object) to the object, and it's up to the object to actually read the data. * Push -- the user reads the data in segments (typically with non-blocking I/O), and pushes it into the object. The Pull API is the easiest to implement, but it has the following problems: * Does not support non-blocking I/O. * Error recovery is complicated (eg if the source is a pipe/socket). * Extra over-head for the common case where you already have read the data. The Push API on the other hand has the following problems: * How to find where to restart the decoding when more data arrives. * How to signal that more data is needed/too much data was received. Suggestions?

2 1

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Pike-devel September 2007