Mailman 3 June 2003 - Twisted

[Twisted-Python] Re: Python 2.3 logging package and Twisted...
by Stuart Hungerford June 24, 2003

June 24, 2003

Itamar wrote: | > In applications using Twisted.web is writing logs with this | > new module the kind of operation that would need to be done in | > a deferred? | | Uh. It can't be *blocking*. Writing to a log file is fine. Writing to a | socket would block, so don't do that. Using a Deferred doesn't magically | make it not block. | | > I understand there's also twisted.python.log, but the issue | > comes up for applications being moved to Twisted that already | > use … [View More]

2 1

[Twisted-Python] IP address, server
by Erik van Blokland June 23, 2003

June 23, 2003

Hello, a humble question for a nudge in the right direction. I'm utterly new to twisted, but I have a lot of experience in python. I'm trying to get a tiny working demo to play with and learn. I've installed twisted 1.0.5 on a Mac G3 running 10.2.6 OSX. Python 2.3b1. runtests seems to work. The tiny http server script works when I take a browser on the same machine as the server and look at http://127.0.0.1:801, I get the expected result, so the server is alive. When I take a look from a … [View More]

2 1

[Twisted-Python] PATCH: doc/historic/index.html reality link
by Lee Harr June 23, 2003

June 23, 2003

>cvs diff -u index.html Index: index.html =================================================================== RCS file: /cvs/Twisted/doc/historic/index.html,v retrieving revision 1.4 diff -u -r1.4 index.html --- index.html 22 Apr 2003 03:51:26 -0000 1.4 +++ index.html 20 Jun 2003 18:25:03 -0000 @@ -100,7 +100,7 @@ we use, and explain how we built tools on top of them which help us make releasing as painless as possible.</p></dd> - <dt><a href="2003/… [View More]

2 1

[Twisted-Python] Minor addition to woven.widgets.ExpandMacro
by Alex Levy June 23, 2003

June 23, 2003

The following diff does two things which are fairly unrelated. One change allows woven.widgets.ExpandMacro to receive a 'macroTemplate' argument instead of requiring a macro's template to reside in a file somewhere. I find this Very Useful(tm). The other is a typo/glitch that I pointed out to Glyph a while back and he hasn't fixed yet. <nudge/><nudge/> ;) Index: twisted/web/woven/widgets.py =================================================================== RCS file: /cvs/… [View More]Twisted/twisted/web/woven/widgets.py,v retrieving revision 1.91 diff -u -r1.91 widgets.py --- twisted/web/woven/widgets.py 23 Jun 2003 07:38:48 -0000 1.91 +++ twisted/web/woven/widgets.py 23 Jun 2003 16:03:48 -0000 @@ -933,17 +933,23 @@ <h3><span slot="greeting" />, <span slot="greetee" />!</h3> </div> """ - def __init__(self, model, macroFile="", macroFileDirectory="", macroName="", **kwargs): + def __init__(self, model, macroTemplate="", macroFile="", macroFileDirectory="", macroName="", **kwargs): + self.macroTemplate = macroTemplate self.macroFile=macroFile self.macroFileDirectory=macroFileDirectory self.macroName=macroName Widget.__init__(self, model, **kwargs) def generate(self, request, node): - templ = view.View( - self.model, - templateFile=self.macroFile, - templateDirectory=self.macroFileDirectory).lookupTemplate(request) + if self.macroTemplate: + templ = view.View( + self.model, + template = self.macroTemplate).lookupTemplate(request) + else: + templ = view.View( + self.model, + templateFile=self.macroFile, + templateDirectory=self.macroFileDirectory).lookupTemplate(request) ## We are going to return the macro node from the metatemplate, ## after replacing any slot= nodes in it with fill-slot= nodes from `node' Index: twisted/world/structfile.py =================================================================== RCS file: /cvs/Twisted/twisted/world/structfile.py,v retrieving revision 1.5 diff -u -r1.5 structfile.py --- twisted/world/structfile.py 8 Jun 2003 11:27:39 -0000 1.5 +++ twisted/world/structfile.py 23 Jun 2003 16:03:48 -0000 @@ -195,7 +195,7 @@ for row in self: f.write('<tr><td>') f.write('</td><td>'.join([str(x) for x in row])) - f.write('</tr>') + f.write('</td></tr>') f.write("</table>") def close(self): -- Alex Levy WWW: http://mesozoic.geecs.org "Never let your sense of morals prevent you from doing what is right." -- Salvor Hardin, Isaac Asimov's _Foundation_ [View Less]

1 0

[Twisted-Python] New Guard question
by jml＠ids.org.au June 23, 2003

June 23, 2003

I've been trying to figure out the new guard and cred, with the goal of perhaps switching Issues over to use guard rather than the custom auth it does now. Looking at the webhappyrealm.py example in the sandbox, it's not very clear what the returned resources represent. Also, returning Resources seems a little weird. To me, a resource is whole page, why would I want a whole page returned? cheers, jml

3 3

[Twisted-Python] FYI: .tap name clash
by Tommi Virtanen June 23, 2003

June 23, 2003

FYI, there's another project using the filename extension ".tap". http://wav-prg.sourceforge.net/ -- :(){ :|:&};:

1 0

[Twisted-Python] twisted.web.client and status 202
by Tommi Virtanen June 22, 2003

June 22, 2003

Hi. In making TouRSSt (http://toursst.sf.net/, soon in a Debian mirror near you), I noticed Slashdot's RSS feed gives status code 202. http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html 10.2.3 202 Accepted The request has been accepted for processing, but the processing has not been completed. The request might or might not eventually be acted upon, as it might be disallowed when processing actually takes place. There is no facility for re-sending a status code from an asynchronous … [View More]

1 0

[Twisted-Python] Twisted.web and adbapi example?
by Morrow, Paul June 22, 2003

June 22, 2003

I've been pouring through the docs and examples, but can't find an example of a .rpy that generates a page containing the results of a database query. How would such a thing be properly architected? Clearly the adbapi would be involved, but then how do I get the query results into the string returned by my resource? Or would I use a deferred instead? Would someone please share a little code snippet (or point me at one)illustrating how that would best be accomplished? Thanks.

4 4

[Twisted-Python] Weirdness in flow
by Pavel Pergamenshchik June 22, 2003

June 22, 2003

Hi. I couldn't catch cce in IRC for while, so here goes: flow._Deferred does not handle errors properly. I cannot find the place where flow._Deferred.errback is called, and it doesn't work anyway. I put a short program that demonstrates the behavior on http://dunce.mine.nu:8080/flow_simplefailure_test.py No matter what I do, I cannot get "got from getPage yield" to get printed or errback to get triggered. Paul

1 0

Re: Fwd: [Twisted-Python] New Guard question
by Glyph Lefkowitz June 21, 2003

June 21, 2003

Begin forwarded message: > From: adalke(a)mindspring.com > Hi Glyph, > > I'm travelling right now, and can't send from the account I use > to send mail to Twisted, so I'm sending this via personal email. > Feel free to CC to Twisted as needed. Your wish is my command. > You gave two examples, a badidea one and a sketch of a new > one. I think you are presenting a straw-man badidea. They were demonstrations of a coding style which I have actually seen several … [View More]examples of. I think you're actually agreeing with me, because I can't find anything in this message that indicates otherwise. Let me be clear on the distinction between these styles: what I am saying is *good* is keeping the avatar in the resource (which the new-cred enabled UsernamePasswordWrapper pretty much requires). This is good because resources are lightweight and it establishes a good encapsulation barrier. Rather than having a Resource which depends on state in the session, or a non-Resource which does resource-like things and accept different arguments. You have a resource that can _always_ render itself as long as it is internally consistent. What I am saying is *bad* is retrieving the avatar from the session every time in your own application code. Now, obviously the code-path to retrieve the avatar object does have to interact with your session, because HTTP sucks, but this is tricky and you should wrap it if you can. Guard does it for you, and I think it does a good job, but if it doesn't you should probably contribute fixes or suggestions for how it could be better, and not write your own session-masher. > Were I to write such code, it would look like > > import weakref > request_info = WeakRef.WeakKeyDictionary() > > class MyGuardResource(Resource): > def __init__(self, guarded_resource, realm): > Resource.__init__(self) > self.realm = realm > self.guarded_resource = guarded_resource > > def getChild(self, name): > # Fix as needed if getChild is supposed to return None for no > named child > return MyGuardResource(self.realm, > self.guarded_resource.getChild(name)) > > def render(self, request): > # negotiate session > s = request.getSession() > if s is None: > return request.setupSession() > > # Logged in? > c = s.getComponent(ICredSession) > if c is None: > request.setHeader('content-type', 'text/plain') > return 'it looks like you are anonymous please log in!' > avatar = c.getLoggedInForRealm(self.realm) > # Either use a weakref like this > request_info[request] = {"avatar": avatar} > return self.guarded_resource.render(request) > # or change the call API for request, like > # return self.guarded_resource.render(request, avatar) In this example, self.guarded_resource would not be a Resource, but some special GuardedResource subclass which produced output for an avatar rather than a request. > Neither the weakref nor the tweaked parameters are > particularly satisfying. I could make it nicer by assuming I > can call the Resource constructor, as you seem to allow. But then, what would you call the Resource constructor with? Perhaps the avatar? :). > It's hard to compare to your "not so bad" example because > you don't show how your two resources are added to the > system. As written, you don't allow either one to have dynamic > children, so that being the case, the following would be a > simpler replacement for your "badidea" code The "bad idea" resource is a resource that is statically added on the "web" side of things by a putChild or a .rpy or somesuch, and thus can only know what realm it is being added for, and must be added in parallel with the guard object. The "not so bad" resource is meant to be instantiated by a Realm. Most likely, adaption to the appropriate interface using the components system will be automated by some common Realm base class when we add one, but for now it's up to you to figure out to return a Resource. In this case, only the UsernamePasswordWrapper has to know about the Portal, and the realm can create an appropriate Resource without knowing anything else about the structure of other web code. Actually the "bad idea" resource can't do dynamic children without doing boilerplate, whereas the "not so bad" resource may implement getChild in a naive way and it will work properly. > class MySimpleGuardResource(Resource): > def __init__(self, realm, authorized, not_authorized): > Resource.__init__(self): > self.realm = realm > .. same for the other two parameter... > def render(self, request): > # negotiate session > ... getSession() .. if is None ... setupSession > c = s.getComponent(ICredSession) > if c is None: > return self.not_authorized().render(session) > return self.authorized(c).render(session) > > in which case I can identically use your not-so-bad > classes. Well, yes, that is the whole point of guard. > So you say: >> Consider, if you want to use woven.guard's nifty session-negotiation >> feature, but also keep your user-specific code in a session, your code >> will look like this: Let me clarify: If you want to use woven.guard's nifty session-negotiation feature, but also implement service-specific functionality in the resource and retrieve it from the session yourself rather than just creating an intelligent stateful resource, your code will look like this: > but I don't see why the latter follows from the former, given that with > a simple adapter it appears to allow your code. ... which is what Guard is doing. The point is that I am encouraging users to use this adapter and have the Realm create an appropriate resource by calling its constructor, rather than attempting to have a resource initialized dumb and then re-customize itself on each call to render(). If you still really think this is a straw-man example of bad style, then I applaud your optimism, but it's seriously a very common mistake :-). After all, most web frameworks have the notion of rendering a page as pretty basic, and then things that are "dynamic" are implemented with do-I-have-a-logged-in-user conditionals, not object references. > Mind you, I'm still trying to understand how to do authentication in > Twisted, either cred or mind style. I know how to do it myself, so > that's my starting point. "cred or mind style"? There are only two styles, one of which is old-cred (Service/Perspective/Authorizer), which is awful and horrible and don't use it please unless you have to interact with code that hasn't been ported from it yet, and new-cred (Realm/Avatar/CredentialsChecker), which will be available in 1.0.6, and all this discussion is about. > Regarding names, IMHO, "Guard" is a better name than "Portal", > and you could use "UsernamePasswordGuard", etc. rather than > "...Wrapper". Hmm. 'Guard' is a different thing from 'Portal', but I think that you're right about UsernamePasswordGuard rather than UsernamePasswordWrapper. [View Less]

1 0