Pike-devel

pike-devel@lists.lysator.liu.se

3044 discussions

Protocols.DNS_SD.Service
by Jonas Walld�n ＠ Pike developers forum 10 Apr '05

10 Apr '05

I just checked in a first attempt at an interface for DNS Service Discovery in Pike 7.7. This lets Pike programs register services so that Rendezvous/Zeroconf-aware applications on the same network can see them. For example, I plan on using it to announce web servers and have them show up as Safari bookmarks automatically. The current implementation is designed to work with Mac OS X 10.3 or later as well as Linux (using libhowl). However, I don't have a Linux machine with the developer package of libhowl installed so I'm sure some tweaks will be needed for it to compile correctly. A natural step forward would be to design a Protocols.DNS_SD.Browser class next. This can be a bit trickier depending on what type of API the high-level code should have (e.g. async callbacks or just a snapshot of the currently available services). If anyone feels like helping out it would be great! (BTW, thanks to Grubba for adding the missing testsuite.in to fix the yellow blupps in Xenofarm.)

3 7

utf8_to_string crippling
by Peter Bortas ＠ Pike developers forum 08 Apr '05

08 Apr '05

I'm not convinced that I like the utf8_to_string crippling. I can see why you don't want to do it in a lot of applications, but how do I decoce the non-shortest form with this change?

5 38

utf8_to_string crippling
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 04 Apr '05

04 Apr '05

Decoding such malformed surrogate sequences does not affect the ability of an application to treat ASCII characters without decoding UTF though, which if I understood correctly was the only reason for the extra stringency in the decoder...

2 1

TypingQuadrant
by Martin Bähr 01 Apr '05

01 Apr '05

hi, i just tried to understand all of what is written here: http://c2.com/cgi/wiki?TypingQuadrant and figure out where pike fits in there. i somehow feel that pike fits neither strong dynamic typing nor strong static typing and then further down it talks about aspects that i don't know how to apply to pike. can you help? greetings, martin.

1 0

Xenofarm
by Martin Nilsson (DivX Inc.) ＠ Pike (-) developers forum 01 Apr '05

01 Apr '05

Xenofarm has falsely reported dmalloc-builds for 7.7 as green for a while since, while dmalloc complains, no error code is returned and no =LEAK= is outputted.

1 0

nessie - the loch ness monster of programming languages
by Martin Bähr 01 Apr '05

01 Apr '05

the loch ness monster is real a new programming language has been sighted, it goes by the affectionate name "nessie" and it claims to be the loch ness monster of programming languages. the main trait of nessie it borrows from python: indenting. the main difference to python: it is typed, something that python developers are now considering for python 3. ever since i started using python myself i have to agree, indenting is one of the best ideas in the evolution of the syntax of programming languages that has been made in a long time. indenting is a must for readable code anyways and anyone who thinks that indenting is a problem is either writing sloppy code or has simply never tried python. i like concise and explicit syntax without redundancy. and that's something that nessie promises to offer. the syntax should serve both the user and the compiler. the compiler should not try to guess what the user wants to do, but it should also not force the user to be overly verbose. nessie offers a number of interesting things that python does not (yet) have. one of them is static typing. this topic has always raised a controversy in the python community. nessie may be the way out for those that do like static typing without forcing their opinion onto the rest of the python community. there is also the very compact use of operators on all kinds of data operations, like splitting a string: instead of s.split(" ") nessie uses s/" " you divide the string at the spaces. the reverse of course also works: arr*" " joins an array to a string. there are many more operators and they can all be applied to almost any type. there is a well evolved runtime library that offers an astonishing set of features from image manipulation, a calendar module that supports any kind of date manipulation you have ever dreamed of, support for most internet protocols, xml, database access, you name it, it's there. like python nessie is object oriented, but it offers class member protection that allows to add security features to your application that are simply not possible in python. one interesting feature here is nessies class syntax. you don't need to mark a class with a keyword, but just but the contents of your class into a file. the filename becomes the class name. instantiate or inherit the file elsewhere and you are done. a bit more about types: nessie is statically typed but it does not make you go out of your way to make things work like java does. there is a mixed type which will hold anything, and brings you back to the same dynamic typing flexibility that python offers. you may also combine any types as you like: string|int|float. incidentially a similar syntax for types is being discussed for python. container types may be specified generic: array (which is really array(mixed)) or more detailed: mapping(string:array(array(int(0..359))). the latter is a dictionary with keys of type string and 2-dimensional arrays with integers ranged from 0 to 359 as values. as you can see the typing is very flexible, you may be as specific as you want and as generic as you need. let's have a look at nessie itself: nessie: void main() write("hello world!\n") pike: void main() { write("hello world!\n"); } python: print "hello world!" nessie is more explicit than python but less redundant than pike. apart from the indenting syntax, nessie is essentially a dialect of pike. it uses the pike runtime which brings with it a great amount of features that have been evolving since the late 80ties. the features described above are really all just inherited from pike. nessie actually just adds the indenting. a minor difference, that may have a large impact on those that don't like braces. python and pike are both great languages, and so is nessie. it's time for some change. Nessie: http://steam.iaeste.at/nessie/ Pike: http://pike.ida.liu.se/ greetings, martin. -- cooperative communication with sTeam - caudium, pike, roxen and unix offering: programming, training and administration - anywhere in the world -- pike programmer travelling and working in europe open-steam.org unix system- bahai.or.at iaeste.(tuwien.ac|or).at administrator (caudium|gotpike).org is.schon.org Martin Bähr http://www.iaeste.or.at/~mbaehr/

1 0

"temp"
by Peter Bortas ＠ Pike developers forum 31 Mar '05

31 Mar '05

Just so that no one gets confused: The Xenofarm results from the host "temp" are compiled in a Cygwin environment. I will fix the hostname eventually.

5 10

cmod-format
by Martin Nilsson (DivX Inc.) ＠ Pike (-) developers forum 30 Mar '05

30 Mar '05

The recent complete backport of precompile.pike from 7.7 to 7.6 breaks some cmod files, since the API has changed for object|void variables. They are no longer svalues but objects. Was this intentionally? If so, I think it was a bad idea.

1 0

Image.JPEG
by M Norrby (rabies), Enea Epact ＠ Pike (-) developers forum 30 Mar '05

30 Mar '05

Nilsson has fixed the configure stuff for Image.JPEG (in 7.7) so that the transform stuff now is included in the Windows build too. I think it is backported to 7.6. I was wondering if someone with the right permissions could backport it to 7.4 if possible. It would be nice if the next Windows version of Roxen (based on 7.4) had the same Image.JPEG support as the Unix versions.

2 1

Pike 7.7, modules -> post_modules
by David Hedbor ＠ Pike developers forum 30 Mar '05

30 Mar '05

What's up with the pike-requiring modules in the modules directory? Specifically I'm thinking about Gmp, Kerberos, Math and _Regexp_PCRE. All these use cmods and require Pike to be available for compilation. Specifically, if modules now require pike, why have post_modules? More correctly, make it so nothing in modules require pike to compile. If needed, compile a lesser version of pike to bootstrap modules as needed. Gmp and Math might be rather important but why on earth would kerberos and _Regexp_PCRE be in modules and not post_modules?

3 3

Jump to page:

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