Pike-devel

pike-devel@lists.lysator.liu.se

3 participants
3048 discussions

Bug in Protocols.HTTP.Session
by David Gourdelier 08 Jun '04

08 Jun '04

Hi, While developping a non blocking HTTP client for Protocols.XMLRPC I went throught a problem with Protocols.HTTP.Session (with Pike 7.6). Here is the testcode: void data_ok(mixed result) { write("result=%O\n", result); } void fail() { write("fail\n"); } int main(int argc, array argv) { object async_client = Protocols.XMLRPC.AsyncClient("http://www.oreillynet.com/meerkat/xml-rpc/server.php"); async_client["system.listMethods"](data_ok, fail); async_client["system.listMethods"](data_ok, fail); return -1; } The first async_client fail sometime (not always) because ->data() from the Request object returned by Protocols.HTTP.Session returned the empty string while the query is valid and the answer from the server ok. However the second call always succeeded. Now if I remove one call to async_client->`[] everything went fine. The code of the class is as follow: class AsyncClient { static object request; static function user_data_ok; static string _url; void create(string|Standards.URI|Protocols.HTTP.Session.SessionURL url) { _url = url; } static void _data_ok() { mixed result; // sometimes request->data() is "" even if query succeed an answer // from HTTP server is ok if(request) result = decode_response(request->data()); user_data_ok(result); } function `[](string call) { return lambda(function data_ok, function fail, mixed ...args) { user_data_ok = data_ok; request = Protocols.HTTP.Session()->async_do_method_url( "POST", _url, 0, encode_call( call, args ), ([ "content-type":"text/xml"]), 0, _data_ok, fail, ({ })); }; } } Example of a failed query: <- "" <- "" <- "POST /meerkat/xml-rpc/server.php HTTP/1.1\r\nConnection: Keep-Alive\r\nContent-length: 112\r\nUser-agent: Mozilla/5.0 (compatible; MSIE 6.0; Pike HTTP client) Pike/7.6.9\r\nContent-type: text/xml\r\nHost: www.oreillynet.com\r\n\r\n<?xml version=\"1.0\"?>\n<methodCall>\n<methodName>system.listMethods</methodName>\n< params>\n</params>\n</methodCall>\n" <- "POST /meerkat/xml-rpc/server.php HTTP/1.1\r\nConnection: Keep-Alive\r\nContent-length: 112\r\nUser-agent: Mozilla/5.0 (compatible; MSIE 6.0; Pike HTTP client) Pike/7.6.9\r\nContent-type: text/xml\r\nHost: www.oreillynet.com\r\n\r\n<?xml version=\"1.0\"?>\n<methodCall>\n<methodName>system.listMethods</methodName>\n< params>\n</params>\n</methodCall>\n" -> 1101 bytes of data ** 448 bytes of header; 640 bytes left in buffer (pos=461) Attempt to index the empty array with 0. /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/XMLRPC.pmod/module.pmod:267: Protocols.XMLRPC->decode("","\n<!DOCTYPE methodResponse [\n <!ELEMENT methodResponse (params|fault)>\n <!ELEMENT fault (value)>\n\n <!ELEMENT params (param*)>\n <!ELEMENT param (value)>\n\n <!ELEMENT value (#PCDA"+[534]) /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/XMLRPC.pmod/module.pmod:98: Protocols.XMLRPC->decode_response("") /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/XMLRPC.pmod/module.pmod:394: Protocols.XMLRPC.AsyncClient()->_data_ok() /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/HTTP.pmod/Session.pike:317: Request("http://www.oreillynet.com/meerkat/xml-rpc/server.php" - connected - 1101 bytes received)->async_data() /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/HTTP.pmod/Query.pike:986: Protocols.HTTP.Session()->SessionQuery(200 OK)->async_fetch(Request("http: //www.oreillynet.com/meerkat/xml-rpc/server.php" - connected - 1101 bytes received)->async_data) /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/HTTP.pmod/Session.pike:295: Request("http://www.oreillynet.com/meerkat/xml-rpc/server.php" - connected - 1101 bytes received)->async_ok(Protocols.HTTP.Session()->SessionQuery(2 00 OK)) /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/HTTP.pmod/Query.pike:135: Protocols.HTTP.Session()->SessionQuery(200 OK)->ponder_answer(0) /usr/local/pike/7.6.9/lib/modules/Protocols.pmod/HTTP.pmod/Query.pike:224: Protocols.HTTP.Session()->SessionQuery(200 OK)->async_read(0,"HTTP/1.1 200 OK\r\nDate: Tue, 08 Jun 2004 15:43:22 GMT\r\nServer: Apache/1.3.29 (Unix) PHP/4.3.4 mod_perl/1.29\r\nP3P: policyref=\"http://www.oreillynet.com/w3c /p3p.xml\",CP=\"CAO DSP COR CURa ADMa DEVa T"+[907]) /usr/local/pike/7.6.9/lib/modules/Stdio.pmod/module.pmod:815: Stdio.File("socket", "208.201.239.37:80", 777 /* fd=11 */)->__stdio_read_c allback() _static_modules.Builtin()->Backend: Pike.Backend(0)->`()() -> 1120 bytes of data ** 467 bytes of header; 640 bytes left in buffer (pos=480) fetch data: 640 bytes needed, got 640, 0 left buf[datapos..] : "<?xml version=\"1.0\"?" buf[..datapos+len] : ">\n</methodResponse>" result=({ /* 1 element */ ({ /* 9 elements */ "meerkat.getChannels", "meerkat.getCategories", "meerkat.getCategoriesBySubstring", "meerkat.getChannelsByCategory", "meerkat.getChannelsBySubstring", "meerkat.getItems", "system.listMethods", "system.methodHelp", "system.methodSignature" }) }) Also I'd like to know if there are practical examples of how to use keep-alive with Protocols.HTTP.Session and how does it work basically (especially I don't understand give_me_connection since it returns a Query object, shouldn't it returned a Request object ?). / David

1 0

Data types in foreach
by Alexander Demenshin 06 Jun '04

06 Jun '04

Hi there, Just found that following construct: mapping m = ([ "key": "value" ]); array a = ({ m }); foreach (a; ; int v) // Note the type of v write("%O", v); ...does work (7.6.6) - it prints mapping, while typeof(v) is int (_typeof(v) is mapping, however). Actually, it doesn't matter which type is used to declare v - it will work anyway - both for indexes and values: foreach (anyvalue; int k; int v) // This will always work, regardless write("%O=%O\n", k, v); // of type of anyvalue and it's elements Seems to be a bug or? Regards, /Al

3 2

Protocols.[SL]MTP.Server and return codes.
by Xavier Beaudouin 04 Jun '04

04 Jun '04

Hello there, I am working on making some SPF/SRS compliant LMTP/SMTP forwarder. I am asking that callbacks (cb_mailfrom / cb_rcptto / cb_data) to return int|array instead of int. Why ? Because some special application can return a message different that "only" "250 Requested mail action okay, completed". For example, I want to add message id inside this 250 after data eg : 250 Requested DATA action okay, message queue id 12120910.2244 Also when rejecting on RCPT TO: I'd like to say why : 550 - SPS mail from invalid : too old Can I do such modification on 7.7 and 7.6 tree as well ? /Xavier -- Xavier Beaudouin - Unix System Administrator & Projects Leader. President of Kazar Organization : http://www.kazar.net/ Please visit http://caudium.net/, home of Caudium & Camas projects

2 2

expected?
by Mirar ＠ Pike developers forum 31 May '04

31 May '04

Is this the expected behaviour when accidentally choosing a directory as the master? % pike -m . /home/mirar/pike/src/object.c:477: Fatal error: low_read_file(.) failed, errno=21 No stack - no backtrace. zsh: abort pike -m .

3 4

FreeType woes
by Fredrik (Naranek) Hubinette (Real Build Master) ＠ Pike (-) developers forum 31 May '04

31 May '04

Anybody here recognize this bug? http://fredrik.hubbe.net/hacker/viewtopic.php?t=162 Cliff notes: Image.Fonts.Font in Pike 7.6.6-2 fails to render images.

3 3

conference thoughts
by Bill Welliver 30 May '04

30 May '04

I was discussing the topic of the next Pike conference with Martin Baehr earlier, and we thought that it might be good to share our thoughts and get a discussion going. Basically, the conversation centered around 2 points: where to have it, and when to have it. On the first point of location, Martin has done some initial legwork. He's located a few potential locations, at least one of which has real potential in my mind. I'll let him talk about them if he feels it appropriate. The subject of when to have the conference is a little bit more important to me, as I'm starting a new job and need to know when I'm going to be away from work. Ideally, I'd like to be able to attend, and for that to happen, a date range somewhere between the middle of September to the beginning of October would be ideal. I say ideal because it gives us about a year since the previous conference, it's not in a peak vacation period for many people with jobs, and the weather hasn't turned unpleasant for those with travel/holiday plans incorporating the conference. For those who aren't keeping track, last year's conference ran roughly between the 24th and 28th of September in Paderborn, Germany. Any thoughts? Bill

11 17

Re: conference thoughts
by Leif Stensson, Lysator ＠ Pike developers forum 29 May '04

29 May '04

> Travel to Sweden is significantly more expensive for those of us > outside of Europe, not to mention hassle and time lost switching > airlines... : ) Of course. But I imagine roughly the same amount of hassle and expense occurs if the conference is in Germany, so I didn't mention that. :-) > Besides, if we're getting picky about location, wasn't the conference > the year before at Roxen HQ? (Not that I'm trying to be difficult or > anything) Actually, no. There were no Pike conferences at all before the one at Linköping University in 2002. There were brief visits at Roxen HQ during the Roxen User Conferences earlier years, but the conferences were not held at Roxen HQ, but rather in a specially rented movie theatre nearby. > One thing might be to consider locations where there are known pockets > of Pike developers that we might want to engage. There are a few in > europe east of germany, so perhaps that's a possibility too. Could be. Also, several of the countries east of Germany (including Poland, Estonia, Latvia, and Lithuania) joined the European Union earlier this month, after years of negotiations, so it has just become easier both for the people there to travel elsewhere in the EU, and for us to get to them. / Leif Stensson, Lysator Previous text: >2004-05-29 04:27: >Subject: Re: conference thoughts >-------------------------------------------------------------------- >Sounds like a good task for a group of his new pike students :) > >> Another possibility might be Dresden. Uwe Aßmann, who sponsored Pike >> development at Linköping University, is moving there this summer to >> start working at Dresden University. (I don't know if he has the time >> to arrange a conference there so soon after getting there, but it's >> a possibility.) >> >>> linköping is not much better however because for many it is to far >>> away and to expensive to get there. >> > >Travel to Sweden is significantly more expensive for those of us >outside of Europe, not to mention hassle and time lost switching >airlines... : ) > >Besides, if we're getting picky about location, wasn't the conference >the year before at Roxen HQ? (Not that I'm trying to be difficult or >anything) > >One thing might be to consider locations where there are known pockets >of Pike developers that we might want to engage. There are a few in >europe east of germany, so perhaps that's a possibility too. > >Bill > > > / Brevbäraren >

1 0

Events in pike?
by Marcus Agehall (Scanian) ＠ Pike (-) developers forum 29 May '04

29 May '04

Is there anyone except for me that likes to use events in their code? In several of my projects, I've implemented eventsystems with a register_event() method and a fire_event() method in a very simple fashion. The thought has occured to me, to write something more general and put it within Pike. But before I spend any time on it, I'd like some input on the matter. For instance, is it useful enough to write put in Pike? Would you like it to work in some special way? How do you normally implement events in Pike?

7 15

Threads under win2k
by Marcus Agehall (Scanian) ＠ Pike (-) developers forum 27 May '04

27 May '04

Using Pike 7.5.17 compiled by Peter Bortas, I got the following error: \home\peter\hack\Pike\7.5\src\threads.c:65: Fatal error: \home\peter\hack\Pike\7.5\src\threads.c:182: WaitForSingleObject((*(&c->lock)), 0xFFFFFFFF) == ((((DWORD )0x00000000L) ) + 0 ) Unexpected error from thread function: 6 Backtrace at time of fatal: Is that a known bug? My code starts a lot of threads (about 20 simultaniously running) which scans directories for their permissions.

2 2

Compiler bug
by David Gourdelier 27 May '04

27 May '04

Hi, Here is a new bug of Pike 7.6(.9) on FreeBSD (5.2.1): $ cat segfault.pike void set_expire_hook(function exphook, void|mixed ...fargs, string|object id) { } int main() { } $ pike segfault.pike segfault.pike:1:Can't define more arguments after ... /home/david/Pike/7.6/src/pike_types.c:717: Fatal error: Unexpected type on stack: 255 (expected 0) Backtrace at time of fatal: /usr/local/pike/7.6.9/lib/master.pike:346: object(/usr/local/pike/7.6.9/lib/master.pike)->compile_string("void set_expire_hook(function exphook, void|mixed ...fargs, string|object id)\n{\n}\n\nint main()\n{\n}\n","/home/david/scripts/segfault.pike",0,/home/david/scripts/segfault,0) /usr/local/pike/7.6.9/lib/master.pike:813: object(/usr/local/pike/7.6.9/lib/master.pike)->low_findprog("/home/david/scripts/segfault",".pike",0,0) /usr/local/pike/7.6.9/lib/master.pike:907: object(/usr/local/pike/7.6.9/lib/master.pike)->findprog("/home/david/scripts/segfault",".pike",0,0) /usr/local/pike/7.6.9/lib/master.pike:943: object(/usr/local/pike/7.6.9/lib/master.pike)->low_cast_to_program("/home/david/scripts/segfault","/usr/local/pike/7.6.9/lib/master.pike",0,0) /usr/local/pike/7.6.9/lib/master.pike:977: object(/usr/local/pike/7.6.9/lib/master.pike)->cast_to_program("/home/david/scripts/segfault.pike","/usr/local/pike/7.6.9/lib/master.pike",0) Abort trap (core dumped) It's 100% reproductible. / David

2 2

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