Pike-devel August 2003

pike-devel@lists.lysator.liu.se

8 participants
25 discussions

Image.SSL?
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 04 May '04

04 May '04

Pike v7.3 release 62 running Hilfe v3.5 (Incremental Pike Frontend) > Image.gurkburk; /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/handshake.pike:978:Index 'decode_certificate' not present in module 'X509'. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:26:Error finding program /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:26:Illegal program pointer. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:35:Illegal program pointer. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:36:Illegal program pointer. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:37:Illegal program pointer. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:41:No inherit or surrounding class alert. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:42:No inherit or surrounding class urgent. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:43:No inherit or surrounding class application. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:46:No inherit or surrounding class handshake. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:123:No inherit or surrounding class alert. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/connection.pike:126:No inherit or surrounding class urgent. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:9:Error finding program /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:9:Illegal program pointer. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:120:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:147:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:364:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:397:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:402:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:410:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:420:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:425:Must return a value for a non-void function. /pike/home/marcus/Pike/7.3/lib/modules/SSL.pmod/sslfile.pike:434:Must return a value for a non-void function. Compiler Error: 1:Index 'gurkburk' not present in module 'Image'. > Two questions: 1) Why is it trying to load SSL.pmod? 2) Why is it failing?

12 137

More about ADTs
by Peta, jo det �r jag ＠ Pike developers forum 04 May '04

04 May '04

I've made som changes and here is a new suggestion for a common interface of the new ADTs. Common API for new ADTs in Pike 1. Introduction This document describes a common platform for new ADTs in pike to stand on. The purpose of the platform is to make the new ADTs as flexible as possible and possible to use together with generic algorithms. The base of the platform is the Collection concept, being the base concepts of all the ADTs, and the Iterator concept making it possible to iterate over collections without knowing the underlying structure. Besides the two base concepts this document also describes a few sub-concepts that are related to the two base concepts according to the two figures below. Figur 1 Collection concepts Collection |-----------> Ordered Collection | | | |--------> Sequence | |--------> Ordered Set <------| | |--------> Ordered Map <--| | |----> Map --------------------------------------| | |----> Set ------------------------------------------| Figur 2 Interface concepts Iterator -------> Ordered Iterator 2. Collection concepts 2.1. Collection Collection is the base concept of all the ADTs. The reason it is called collection is that all the ADTs in the library is in some way a collection of items. To implement a collection the following methods must be implemented iterator _get_iterator() Return an iterator for this ADT positioned at the first item in the ADT. int _sizeof() Return the current size of the ADT. Return -1 if the size could not be decided or is truly unlimited. array _values() Return an array with the values in the ADT. int _has_value(mixed value) Returns 1 if value is in the value domain of collection, or 0 (zero) if not found int _equal(mixed coll) Decide if the collection coll and this collection contains the same elements, an if it is important, in the same order. string _sprintf(int conversion_type, mapping(string:int)|void params) Present the collection in a human readable way. For more information see _sprintf in the pike manual. int is_empty() Return 0 if the ADT do has elements. int max_size() Return the maximal size of the ADT if there is any, otherwise -1. void clear() Clear the ADT. iterator first() Return an iterator for this ADT positioned at the first item in the ADT, same as _get_iterator(). iterator last() Return an iterator for this ADT positioned at the last item in the ADT. 2.2. Set Set is a collection containing any number of items but only one of each item. It is a sub-concept of Collection and must thereby besides implementing the methods below also implements the methods in Collections. int '[](mixed item) Return 1 if the item could be found in the set, 0 if it couldn't int '[]=(mixed item, int exist) Add the item to the set if exist is true remove the item if exist is false. iterator get_iterator(mixed item) Get an iterator for this ADT positioned at the item or null if the item could not be found int insert(mixed item) Add the item to the set if it isn't a part of the set. Return true if the item was added, else false. mixed delete(mixed item) Remove the item from the set. If the item was removed it will be returned otherwise a 0 zero_type is returned. multiset cast("multiset") The set should at least be able to cast to a multiset, additional castings are optional. 2.3. Map A map is a data structure that allows you to associate a key with one or more values and then quickly retrieve those values using the key. To implement a Map the ADT must have all the methods in this concept and in the Collection concept. mixed '[](mixed key) Return the item associated with the key if it could be found, 0 if it couldn't mixed '[]=(mixed key, mixed item) If there is an item with the key key then the associated item will be overwritten with the supplied item. If there is no key in the map a new key-item pair will be added. iterator get_iterator(mixed key) Get an iterator for this ADT positioned at the key or null if the key could not be found. mixed insert(mixed key, mixed item) If the key doesn't exist or duplicates are allowed, associate the value with the key and return 0, otherwise don't modify the map and return the current value associated with the key. mixed delete(mixed key) Remove the pair with the key key. Return the keys value if it was removed, if the key could not be removed return zero_type. mapping cast("mapping") The map should at leas be able to cast to a mapping, additional castings are optional. 2.4. OrderedCollection An OrderedCollection is a collection that has indices that could be used to find the values and the indices are strictly ordered is some way. Any type could be used as an index if it satisfies a few rules. See Index for more information. There is also nothing that prevents the use of the values as indices as long as the value satisfies the rules for an Index. int _has_index(int index) Returns 1 if index is in the index domain of collection, or 0 (zero) if not found array _indices() Return an array with the indices of the collection. 2.5. Sequence A sequence is a sub-concept of an Ordered Collection. It is a linear collection that allows index-based access and automatically expands to accommodate new elements. A vector is a good example. mixed '[](int ind) Return the item at index ind in the sequence. Throw an error if ind is out of range. mixed '[]=(int ind, mixed item) Set the index ind to be item. Throw an error if ind is out of range int _search(mixed item, void|int start) Return the index of the first item in the sequence with the index start or higher that match the item item. -1 if no object was found. Throw an error if start is out of range. void insert(index ind, mixed item) Insert an item in the sequence at the index ind. The size of the sequence is increased by one and the items at index and above have their indices increases by one. Throw an error if ind is out of range. int(0..1) delete(mixed item) Remove the first value that are equal to item from the sequence. Return 1 if a item was removed, otherwise 0; void add(mixed item) Add the object item as fast as possible to the sequence. int delete_at(int ind) Remove the item at the index ind. The size of the sequence is decreased by one and the items above ind have their indices decreases by one. Return 0 if the removal failed. Throw an error if ind is out of range. iterator get_iterator(int ind) Get an iterator for this ADT positioned at the index ind. Throw an error if ind is out of range. array cast("array") The sequence should at least be able to cast to an array, additional castings are optional. 2.6. OrderedMap An OrderedMap is a map where the keys are orderable and the iterator belonging to an OrderedMap will traverse the values in that order. 2.7. OrderedSet An OrderedSet is a set were the values are orderable and the iterator belonging to an OrderedSet will traverse the values in that order. 3. Iterator concepts 3.1. Iterator The iterator concept does not differ much from the present day iterator. Only a few new methods are added to make it possible to step backwards and compare iterators. int(01) '!() Should return 0 (zero) when not at end of collection, and 1 at end of collection void '+=(int n) Should advance the specified number of steps iterator '-=(int n) Move the iterator n steps backwards int '==(iterator iter) Return true if the two iterators is pointing to the same object. Throw an error if the two iterators belong to different collections. mixed index() Should return the current index. mixed value() Should return the current value int has_previous() Return true if the iterator can move one step backwards. int has_next() Return true if the iterator can move one step forwards. It have the same use as '!(). collection get_collection() Return the collection this iterator iterates over. Optional void set_value(mixed value) Set the value of this iterator. This method should only be available if it is possible to set the value via an iterator. Optional void add(mixed value) Add a new item to the collection after the item the iterator currently is pointing at. This method should only be available if it is possible to add entries via the iterator Optional mixed remove() Remove the item the iterator currently is pointing at and move to the next item. This method should only be available if it is possible to remove entries via the iterator 3.2. OrderedIterator An OrderedIterator is an iterator working on an OrderedCollection. It has two characteristics, first it traverse the collection in the same order each time it is traversed, if the there hasn't been any changes in the collection, beginning with the item with the lowest index and ending with the item with the highest index. Secondly it is possibly to say that an iterator is less then or greater then another iterator. int(0..1) '<(OrderedIterator iter) Return 1 if iter is positioned after this iterator, otherwise 0. Throw an error if the two iterators belong to different collections. Int(0..1) '>(OrderedIterator iter) Return 1 if iter is positioned before this iterator otherwise 0. Throw an error if the two iterators belong to different collections. 4. Index If a type could be used as an index in an OrderedCollection then it must be possible to use the following operations on it. '== If the two indexes a and b is equal it should return 1 otherwise 0. The following rules must also be satisfied. a == a must be true a == b implies b == a !(a == b) implies a != b a == b and b == c implies a == c '< If the index a is ordered lower than b it should return 1 otherwise 0. The following rules must also be satisfied. a < a must be false a < b implies !(b < a) a < b implies (b > a) a < b and b < c implies a < c '> If the index a is ordered higher than b it should return 1 otherwise 0. The following rules must also be satisfied. a > a must be false a > b implies !(b > a) a > b implies b < a a > b and b > c implies a > c

11 73

SDL
by David Hedbor ＠ Pike developers forum 04 May '04

04 May '04

I have now implemented and checked in SDL.Surface->set_pixel and get_pixel methods. Note that they do no boundary checking so if you write outside the surface... Well, go figure. I could add it but I'd rather avoid any checks I can for performance reasons...

6 18

Starting new pike processes from the testsuite
by Niels M�ller (ny flexiblare br�stkorg) ＠ Pike (-) developers forum 24 Feb '04

24 Feb '04

What's the recommended way of starting separate processes (which are Pike programs) from the testsuite? I want to make a test case for the SSL code, by starting a https process using the SSL.pmod/https.pike program (which I'm not sure it really belongs in the module tree) or some similar code, and then use the client code in the url/http modules to connect to it.

9 35

Gentoo Linux ebuilds for pike-7.4.20 available
by Rainer Groesslinger 01 Sep '03

01 Sep '03

Hello guys, first, I am new to the list, just subscribed some hours ago. Well, I am also new to pike, I heard of pike for the first time yesterday but I have to admit that it looks really promising and I'll have a closer look at it ;) I have committed an ebuild for pike-7.4.20 to gentoo's portage tree. Gentoo users can install it just like every other ebuild, too. For reference, you can also take a look at it using our viewcvs: http://www.gentoo.org/cgi-bin/viewcvs.cgi/dev-lang/pike/ This is the initial commit of a pike ebuild so I'd be warmly interested in every sucess and failure story regarding the ebuild ;-) With your feedback I hope to steadily improve the pike ebuilds and get it into the stable tree quite soon. -- Rainer Groesslinger http://dev.gentoo.org/~scandium/

7 21

Gentoo Linux ebuilds for pike-7.4.20 available
by Johan Sundstr�m (p� s�ljakt) ＠ Pike (-) developers forum 31 Aug '03

31 Aug '03

http://www.gentoo.org/cgi-bin/viewcvs.cgi/dev-lang/pike/pike-7.4.20.ebuild?…

1 0

Problems with Backtrace
by kaylus ＠ Pike developers forum 29 Aug '03

29 Aug '03

Pardon my ignorance, but here it goes. I am at the moment writing a general LPC-type server in Pike, mainly for my own amusement, and have run into a problem. First let me give a rundown. I'm sorry if it's long-winded, I like to put all the information out so I don't confuse with my incorrect jargon and don't miss anything important. The server sets up all the constants (located in inherited file) binds to a port and listens. Once a connection attempt is found it sends it to a socket class. The socket class then attempts to clone the player object, (compiles the program pointed to by a #define) and calls the logon() in the compiled object. The server itself, handling connections, cloning of objects, and basically everything works just fine. An LPC-alike clone mud can be built. I have a constant: add_constant("this_user", function_object(backtrace()[-2]->fun)); Which works wonderfully enough while it is inside the player program. My problem: I have a "bin" system for the player commands. It matches the input, searches for an object compiled under that name (i.e. /lib/cmds/who.p when they type who), compiling if not, then calls the do_cmd() function within it. If the this_user() constant is called from within the called command function it returns 0. A backtrace returns an array of 3 Zeroes. The "function_object(backtrace()[-2]->fun)" works fine within it, only the constant has a problem. Currently I am circumventing this by using a #define for the backtrace, but it is rather annoying and unsightly. I was wondering if there was some other way of doing this, or if I was doing something hideously wrong. At least some way of hitting the backtrace info from within such an above example, I can always set checking to make sure it's hitting the right backtrace entry, just need the info first *grin*. Kaylus p.s. Sorry if this was somewhat lengthy, pointless, blather that is hard to read. It's 2am and i've been trying to figure it out for quite some time. Please ignore slight errors in the message!

3 8

Compile-error CASE_INFO_SHIFT0_HIGH
by Martin Baehr 27 Aug '03

27 Aug '03

when trying to compile the released 7.4.20 pike source on a mandrake system, i get the following error: system info: Mandrake Linux release 9.0 (dolphin) for i586 gcc version 3.2 (Mandrake Linux 9.0 3.2-1mdk) libc.so.6 -> libc-2.2.5.so libpthread.so.0 -> libpthread-0.9.so make output: (sorry for the german errormessages) Compiling /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c: In function `find_ci_shift0': /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: `CASE_INFO_SHIFT0_HIGH' undeclared (first use in this function) /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: (Each undeclared identifier is reported only once /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: for each function it appears in.) WARNING: Compiler failure! Trying without optimization! cc1: warning: -Wuninitialized is not supported without -O /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c: In function `find_ci_shift0': /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: `CASE_INFO_SHIFT0_HIGH' undeclared (first use in this function) /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: (Each undeclared identifier is reported only once /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c:287: for each function it appears in.) Compilation command was: /opt/pike-7.4/pike.src/Pike-v7.4.20/build/linux-2.4.19-16mdk-i686/smartlink gcc -DPIKE_SRC_ROOT=/opt/pike-7.4/pike.src/Pike-v7.4.20 -I. -I/opt/pike-7.4/pike.src/Pike-v7.4.20/src -I/usr/X11R6/include -g -mcpu=i686 -I/usr/X11R6/include -W -Wall -Wno-unused -Wcomment -Wformat -Wimplicit-function-declaration -Wmultichar -Wswitch -Wuninitialized -Wpointer-arith -Wchar-subscripts -Wno-long-long -c /opt/pike-7.4/pike.src/Pike-v7.4.20/src/builtin_functions.c -o builtin_functions.o make[3]: *** [builtin_functions.o] Fehler 1 make[3]: Verlassen des Verzeichnisses Verzeichnis =BB/opt/pike-7.4/pike.src/Pike-v7.4.20/build/linux-2.4.19-16mdk-i686=AB make[2]: *** [pike] Fehler 2 make[2]: Verlassen des Verzeichnisses Verzeichnis =BB/opt/pike-7.4/pike.src/Pike-v7.4.20/build/linux-2.4.19-16mdk-i686=AB make[1]: *** [all] Fehler 2 make[1]: Verlassen des Verzeichnisses Verzeichnis =BB/opt/pike-7.4/pike.src/Pike-v7.4.20/build/linux-2.4.19-16mdk-i686=AB make: *** [compile] Fehler 2 greetings, martin. -- Pike Conference 2003 - Sep 25-27 - http://pike.ida.liu.se/conferences/2003/

2 2

module build/install procedure
by Bill Welliver 23 Aug '03

23 Aug '03

I've been fiddling with the existing build tools, as a precursor to a proposal for a module repository ala CPAN, and I've got a few observations: 1. It appears that as far as pike is concerned, C modules can reside anywhere in the module tree, as long as they're a .so file. 2. The dynamic module makefile, however, appears to have no provision for making modules that don't live at the root of the namespace 3. Similarly, the dynamic module makefile doesn't seem to want to deal with modules that either a) have no compiled module component, or b) have more than 1 pike file component, which is assumed to be called module.pmod.in and ends up as ModuleName.pmod, derived from the directory name that is being built in. Can anyone comment on any of these statements? Would it be good to remove these limitations, or perhaps add an external_dynamic_module_makefile that does that? Ideally, it would be good to have a method that handles these variations on the standard pike-distribution-module theme. Also, I've got some other thoughts that anyone is welcome to comment on: 1. We should have a way to insert information into the modref when a new module is installed. This way, if I add a new module, the inline documentation is integrated with the existing modref. Now, I'm not very knowledgable on how this works, but I'm guessing that it would probably involve both installing documentation as part of the build procedure, along with some of the source [xml] files that would be needed to rebuild the documentation after inserting the new material. 2. In order to do an integrated module install/remove feature, we probably need a "receipts" store to keep track of what files belong to with modules, so we can be sure that we delete all of a module, or don't overwrite portions another module accidentally. Part of this could probably be accomplished by making a wrapper to mkdir,install,etc that added or removed entries from a contents file ala the solaris package manager. Obviously, a CPAN for pike [with auto download-build-install] is really just icing on the cake. It's something that we really can't do without having a seamless module management experience, both for users and developers. I'm willing to put some time in on this, but I am hesitant to go mucking around with some of this stuff without getting a blessing from the full-time developers. As always, comments and suggestions are welcome! Bill

4 4

Re: [caudium-devel] Pike conference
by Bertrand LUPART 20 Aug '03

20 Aug '03

>>> CIF rewrite >> IIRC, Grendel began that. > Right, but as I said many times - I'm not going to be able to finish > that > alone. It's enough work to fill 8h/day for 6 months for one person. True, but we may organize for that. >> What more feature people want? > I have some ideas I'd like to present during some discussion on the > conference. (for example object-oriented approach to HTML/RXML/XML - > presentation of the web pages to the caudium modules/scripts as a > collection > of objects on which operations can be performed to alter their > content/properties; Rewriting the Caudium core to separate it from > Pike /in > C, probably/ so that Pike runs in a separate process as a "script > server" > accessible either locally via various IPC means or remotely with > XML-RPC or > whatever protocol seems to fit) That's beyond the CIF :) But sounds really cool. >> I'd change (read: allow to change/make default) the select multiple to >> checkboxes, since many people aren't familiar with these. > What will you do if you have 100 options? (for example 100 modules) > Paging > doesn't seem to be a good idea, and a selection page 10 screens long > is even > worse... I wasn't thinking about the add module page, but modules in general. 100 checkbox are not worse than the add module page in standard mode :) Off course, each one could choose that. > All in all, the look of the CIF is the smallest concern. The beast > sleeps > beneath the surface. And it's a huge beast. Enormous. Yep, but we may be aware of some kind of functionnality before coding things. Could you explain breifly how you started to recode this? >>> RXML parser, and the future of RXML >> >> Grendel has a piece of code for the new RXML parser. > Yep, and some (unpublished) code for a new HTML/RXML parser for code > that > isn't XML-compliant. Thanx, my webmaster will be happy. >> - parser speed (read: Martin's RXML Vs PHP bench) >> - enhancement of do_output_tag() > you wanted to say, reimplementation+redesign, I hope? :) I meant something reliable ;) >> Something allready exists on riverweb. Could we steal some code? > I think it's about quite a different thing. The repository should be > accessible not only from the web with mouse clicks, it should mainly > be a > way to organize the modules/extensions to pike/caudium together with > their > docs so that it's easy to access/find any of those in a canonical way. _nice_ >> It should be integrated into caudium's website for people not having >> CVS >> write access to share their code. > Yep, I agree. There should be a single Caudium module (probably using > another single, Pike, module) that would present a view of the > repository. Could the CIF benefit of these functionalities? >>> Marketing / popularity enhancement >> >> What kind of people do we want to target? Coders? Web-literate >> webmasters? >> Newbies? > The most important target would be the ISPs, I think. Once you get > Caudium > in there, you will slowly win the others. That's because it's useless > to > have an army of drooling webmasters/coders without ISP where they could > install Caudium. That's egg and chicken problem. Would ISPs install Caudium if noone care to use it? Is there any hosting service that provides Caudium hosting except kazar.net? (and linkeo, but that's a different thing) Another target is people that manage their servers. >> It's hard for people to hear about Caudium stuff, coz everything is >> talked >> into mailing lists you have to subscribe to. > Well, Bertrand, I might surprise you but... this is internet, you know > :) - > most of discussions take place on mailing lists/usenet forums :) I may not have been well understood. The difference between mailing list and usenet forums is that you don't have to subscribe to: you just have to select any group, lurk, easily read past message, make a search on it with google (easy to follow a thread), no more read it if you want w/o unsubscribing... Off course, you have archives for mailing list, but never seen a single one that present threads in a usable way. IIRC, our mailing list archives is protected against spammers, ie no search engine would archive that (trying to verify that since yesterday evening, but listes.oav.net seems down) That's why i asked some time ago if we couldn't sync the mailing list and a usenet server. Then we talked about which group, coz we wanted a good group. Then the better killed the best and we didn't come to a solution :) > But I guess > a web-based forum would be a good thing for Caudium to have (another > thing > that would have to be written, I guess) I know people like this kind of stuff. I definitly thinks that this kind of things sux. If someone wants to write a forum, abstraction interface for data is planned for CAMAS 1.3: it could connect to NNTP (or even a database for a full forum behaviour) >> Then, writing tutorials, docs and packaging stuff is mandatory if we >> want >> new people to come. For example, i'm very responsible for allmost 0 >> CAMAS >> layout docs and darwin packaging. > I guess Caudium needs a rewrite first, docs then. And before you > protest, > think about it... [snip] Why do you think i'll protest? :) I fully agree, but docs means tutorials or sample code. I was told by a newbie yesterday that caudium website hardly lacks sample code. > Somebody who is willing to inform "press" about events in the Caudium > community, somebody to arrange for Caudium articles to be published > (s/he > wouldn't necessarily write the articles herself, of course), Press is a good medium. Dunno if that's a coincidence, but look at the unique host visits of caudium.net on march. This was when vida's article made it in the french press. > to contact the > various ad companies which work on the link exchange basis (and, yes, > linkexchange.com is Microsoft, but they should be interested in > exchanging > banners with a site that has PR6 in Google); Is that efficient? I personnaly never nor click nor even look at banner contents. > and the other person would be > responsible for, let's call them, milestones in the caudium > development, for > coordination of various efforts, task assignment - i.e. somebody to > fulfill > the goals set by the project leader in their greater detail. That would be a great thing. -- Bertrand Lupart C'est au pied du mur qu'on voit le mieux le mur -- Bertrand Lupart Recursion: n,m See: Recursion

1 0

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 August 2003