.../zope/interface/interface.py:548: UserWarning: Hashing uninitialized InterfaceClass instance
  warnings.warn('Hashing uninitialized InterfaceClass instance')

This happens because importing nevow imports formless, and formless constructs its own InterfaceClass instances, and then uses them as keys in a dictionary, and doesn't give them __name__ and __module__ attributes.

#36 made 0.11.0 a bit of a dud. Try again with that issue fixed.

Doing this by staring at git log at release time isn't fun.

The code in Athena/init.js at line 1130 detaches only every other client

for (var i = 0; i < self.childWidgets.length; ++i) {
self.childWidgets[i]._athenaDetachClient();

Possible fix by creating a copy of the widgets list could be:

    var childWidgets = self.childWidgets.slice();   //copy self.childWidgets, it will be modified by removeChildWidget

    for (var i = 0; i < self.childWidgets.length; ++i)  {
        childWidgets[i] = self.childWidgets[i];
    }
    for (var i = 0; i < childWidgets.length; ++i)  {
        childWidgets[i]._athenaDetachClient();

This is a bug for which I submitted the above fix and necessary test when divmod.org was still active
How do I proceed?

Thxs, Werner

Imported from Launchpad using lp2gh.

• date created: 2012-02-05T15:20:10Z
• owner: werner-thieprojects
• the launchpad url was https://bugs.launchpad.net/bugs/927179

If the initial transport request contains messages (eg. remote calls), these will be processed before connectionMade is invoke, resulting in inconsistent ordering of events (you might have an exposed method called before any initialization in connectionMade has happened).

Currently the suggested way of defining methods is:

Divmod.Class.subclass(Some.Namespace, 'ClassName').methods(
  function setAttribute(self, some, arguments) {
    // Do some work.
  }
);

Unfortunately earlier versions of Internet Explorer's JScript implementation have the annoying property of leaking named function expressions into the enclosing scope, which in this case happens to be the global one. See http://kangax.github.io/nfe/#jscript-bugs for more horrors.

This became apparent after a long debugging session where the Nevow browser platform implementations define a setAttribute method which leaks into the global namespace becoming window.setAttribute and in turn interfering with the operation of some modern Javascript libraries.

A solution we used in a private fork was introducing an API like:

Divmod.Class.subclass(Some.Namespace, 'ClassName').namedMethods({
  setAttribute: function(self, some, arguments) {
    // Do some work.
  }
});

python-versioneer removes the need to import version information in setup.py from the python package being packaged (something which fails in many contemporary build/install environments). It also avoids the need to specify the version information in more than one place.

Sometimes it can be convenient to pass arguments to the callable in assertThrows directly, instead of creating an anonymous inner function; this would match assertRaises in Python. Also, JSHint warns about creating functions inside loops.

Imported from Launchpad using lp2gh.

• date created: 2011-08-23T10:42:24Z
• owner: jjacobs
• assignee: jjacobs
• the launchpad url was https://bugs.launchpad.net/bugs/831895

Clicking "Code" and then on the only available source link, there's a page which says "import invalid" and "This branch is an import of the Subversion branch from http://divmod.org/svn/Divmod/trunk/Nevow." - but that link is a 404.

There is a copy of the source at https://pypi.python.org/pypi/Nevow , but it lacks revision control history.

Imported from Launchpad using lp2gh.

• date created: 2013-07-22T19:35:23Z
• owner: daira
• the launchpad url was https://bugs.launchpad.net/bugs/1203854

When I enable deprecation warnings in my tests (Tahoe-LAFS, using Nevow-0.11.1 and Twisted > 13.2), I get a warning from nevow:

.../Nevow-0.11.1-py2.7.egg/nevow/appserver.py:129: DeprecationWarning: twisted.web.http.Request.received_headers was deprecated in Twisted 13.2.0: Please use twisted.web.http.Request.requestHeaders instead.

It looks like nevow.appserver.NevowRequest.process is doing:

    self.fields = cgi.FieldStorage(self.content, self.received_headers,
                                   environ={'REQUEST_METHOD': 'POST'})

and Twisted doesn't like that anymore. received_headers provided a dict, which was lossy but convenient for cgi.FieldStorage. requestHeaders is not a dict, and will probably need some new code to transform it into a form that cgi.FieldStorage likes.

While investigating https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2312 , I found that Nevow-0.11.1 plus Twisted-13.0.0 plus python-2.6 can provoke an annoying DeprecationWarning, if you use a nevow.rend.Page with addSlash = True and then render the page in a way that forces a redirect to self-plus-slash.

It looks like twisted.web.http.Request.setHeader() used to accept anything string-able, but now insists upon actual "bytes" objects (probably to avoid confusing unicode errors), and emits a DeprecationWarning if it gets otherwise. We had one place in Tahoe where we used req.setHeader("content-length", len(body)), which ran afoul of this.

But there's another one in nevow itself, in nevow.rend.Page._renderHTTP, where the addSlash behavior is implemented with:

request.redirect(request.URLPath().child(''))

That passes a URL object into setHeader("location", url), which triggers the same warning. I haven't tried making a short proof-of-concept, or tried running Nevow's tests against py2.6.

We don't see this when testing under python2.7, probably because 2.7 does something foolish like disabling all DeprecationWarnings. I haven't done exhaustive testing, but if that's really true, I suspect it'll be a big problem for libraries that want to signal their own deprecations to higher-layer apps (Twisted uses DeprecationWarnings extensively).

I think the fix will be to replace that code in _renderHTTP with something like:

request.redirect(str(request.URLPath().child('')))

This looks like a good patch:

http://bazaar.launchpad.net/~divmod-dev/divmod.org/build-nevow-without-importing-2699/revision/2601

But why do the execfile hack only when setuptools is present? Why not do that same thing in both cases?

Imported from Launchpad using lp2gh.

• date created: 2011-07-18T20:27:58Z
• owner: zooko
• assignee: zooko
• the launchpad url was https://bugs.launchpad.net/bugs/812537

The current travis-ci configuration does pip install --editable ....

This means nothing is really installed. It's just a trick with a pth file or something.

We want Nevow to actually be installable. Probably from an sdist or a wheel. Build these and install them and run the tests against the result. On travis.

The problem is that Nevow imports nevow._version in setup.py which imports twisted. In the setuptools world Nevow should put "twisted" in setup_requires, but setup_requires doesn't actually work with pip because the first thing pip does on downloading is run setup.py egg_info which doesn't work if Twisted is not installed...

Imported from Launchpad using lp2gh.

• date created: 2012-12-17T01:42:40Z
• owner: mwhudson
• the launchpad url was https://bugs.launchpad.net/bugs/1091055

Continuous integration is good.

Binary package hint: python-setuptools

This bug in Debian sid:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=475440

Has now been added to Ubuntu, in Hardy. I confirmed that the bug stops happening when you uninstall the python-setuptools package that comes with Ubuntu Hardy and install the same version of python-setuptools from upstream then the problem does not happen.

Imported from Launchpad using lp2gh.

• date created: 2008-08-22T00:00:10Z
• owner: zooko
• the launchpad url was https://bugs.launchpad.net/bugs/254035

There is currently no way to access the event object from an Athena event handler. Quirksmode has a lot of information [1] that can be useful for implementing this in the "inline event handler" model that Athena currently employs.

It would be worth wrapping the event objects to expose a useful subset of functionality in a cross-browser way.

[1] http://www.quirksmode.org/js/events_access.html#link6

Imported from Launchpad using lp2gh.

• date created: 2011-07-06T15:56:02Z
• owner: jjacobs
• assignee: jjacobs
• the launchpad url was https://bugs.launchpad.net/bugs/806545

What information is this logged message supposed to convey?

Looking at the code it seems like there are some cases where an old message is trying to be retransmitted but I don't fully understand why this would occur. Is this just normal operation or is it an issue that should be resolved?

Either way, currently our logs are absolutely flooded with this message and it doesn't provide any benefit. Can it be removed or resolved, if it is in fact a bug?

There is some out of date information in setup.py now.

(This is copied from the mailing list. I enclose a patch, which includes a unit test.)

J.P's excellent article http://jcalderone.livejournal.com/50890.html
explains how to avoid calling Request.finish() on a request after its
connection was lost, because that now raises an exception. The code in
the article is able to avoid calling finish() because it handles the
rendering itself. The call to finish() is right there in _delayedRender.
But I'm using Nevow (because it's awesome) and all that stuff is done
behind the scenes, where I can't get to it.

My application has a form, which uses the Post/Redirect/Get pattern
(http://en.wikipedia.org/wiki/Post/Redirect/Get) to avoid
double-posting. Because processing is not instant, the browser receives
the HTTP header telling it to redirect before the processing has
finished. When the processing finishes, the Deferred embedded in the web
page fires with a value, Stan finishes flattening the page, and Nevow
calls Request.finish(). By then the browser has closed the connection
and sent in the GET request. So I get an error every single time.
Any page that uses Deferreds and redirects will do as well.

This is a completely self-contained example, to be run with twistd:

from nevow import tags, inevow, loaders, appserver
from nevow.rend import Page
from nevow.url import URL
from twisted.internet.task import deferLater
from twisted.internet import reactor
from twisted.application import service
from twisted.application.internet import TCPServer
from twisted.web import http

class Redirector(Page):

count = 0
message = 0

def _increment(self):
    self.count += 1
    return self.count

def slow_operation(self):
    return deferLater(reactor, 3, self._increment)

def beforeRender(self, ctx):
    req = inevow.IRequest(ctx)
    if req.method == 'POST':
        oldurl = URL.fromContext(ctx)
        newurl = oldurl.remove('cmd')
        req.setResponseCode(http.SEE_OTHER)
        req.setHeader("Location", newurl)
        self.message = self.slow_operation()

def render_message(self, ctx, data):
    return self.message

docFactory = loaders.stan(
    tags.html[
        tags.head[tags.title['Slow page']],
        tags.body[
            tags.h1['Problem with interrupted connections'],
            tags.form(method='post', action='')[
                tags.button(type='submit', name='cmd',
                value='click')[
                    'Click'
                    ]
                ],
            tags.p(render=tags.directive('message'))
            ]
        ]
    )

application = service.Application('Demo')

site = appserver.NevowSite(Redirector())
webservice = TCPServer(8123, site)
webservice.setName('WUI')
webservice.setServiceParent(application)

Imported from Launchpad using lp2gh.

• date created: 2011-11-20T16:08:57Z
• owner: peter-westlake
• the launchpad url was https://bugs.launchpad.net/bugs/892794

These were deprecated in 0.8, which was a long long time ago; also, they're really horrible, and nobody should use them ever. I think it's time to finally remove them.

Imported from Launchpad using lp2gh.

• date created: 2011-08-21T03:29:09Z
• owner: mithrandi
• assignee: mithrandi
• the launchpad url was https://bugs.launchpad.net/bugs/830361

Against Twisted 14.0.0:

Traceback (most recent call last):
  File "/home/travis/build/twisted/nevow/nevow/test/test_appserver.py", line 187, in test_oldStyle
    proto = self.renderResource('/foo')
  File "/home/travis/build/twisted/nevow/nevow/test/test_appserver.py", line 170, in renderResource
    assert proto.transport.disconnecting
exceptions.AssertionError: 

We could use tox for this; another project I work on, Mimic, does it like this: https://github.com/rackerlabs/mimic/blob/master/tox.ini

Lore is no longer maintained.

In Javascript the natural return value of a function is undefined but the Nevow JSON parser does not understand undefined and often this comes up as an issue where a remote call chain ends without needing to specify a return value.

It's not terribly convenient to add return null everywhere to begin with but causing a server-side error, which is then sent to the client and subsequently disconnects them with a rather obscure traceback, is not terribly useful behaviour.

The JSON parser might as well treat undefined the same way as null.

This was accidentally removed along with livepage, but it's really an Athena thing, not a livepage thing, and Mantissa actually uses it. We can deprecate it, though.

nevow.livepage is documented as "Previous generation Nevow Comet support. Do not use this module.". It has seen no significant maintenance since 2006. It is redundant with nevow.athena at least. It may be broken (#48) and misleading to potential users.

It hasn't actually be deprecated in the sense that it emits a deprecation warning if you use it. I think the warning in the module docstring and the narrow audience of Nevow may mean it's okay to just remove it.

As per #57 this is of no value to Athena users.

Apparently this module is left out. I suppose find_packages doesn't find it?

When the Athena message transport is stopped self.requests is set to null which causes breakage in several places expecting to be able to iterate it. The reason these iterations are happening is usually because the transport changed state half way through the processing of a request.

As the warning says, we should be using TestCase.flushLoggedErrors instead.

Imported from Launchpad using lp2gh.

• date created: 2011-08-21T03:29:47Z
• owner: mithrandi
• assignee: mithrandi
• the launchpad url was https://bugs.launchpad.net/bugs/830362

The tests have been failing against Twisted trunk for some time now, but I guess nobody noticed.

https://travis-ci.org/twisted/nevow/builds/142907355

I haven't had a chance to dig into this, but a cursory glance at the failures suggests this is just a problem with our test fakes (eg. FakeHTTPChannel) being too fake for twisted.web to work anymore.

Dear folks:

According to

Do you know if there's any reason it wouldn't work for us to use Nevow 0.11.1 with Twisted == 11.0.0, == 11.1.0, == 12.0.0, == 12.1.0? Those are versions of Twisted that our users are currently able to install on Windows: tahoe-lafs/tahoe-lafs@d888b28 (due to https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2028 (Twisted endpoints introduce a dependency on pywin32) and http://twistedmatrix.com/trac/ticket/6032 (make twisted.internet.endpoints importable on Windows when pywin32 is not installed)).

The requirement Nevow imposes for Twisted >= 13.0 is also causing a different install problem for a user on Mac OS X: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2249# OSX: Installed distribution Twisted 12.2.0 conflicts with requirement twisted>=13.0

After installing python-nevow, the "trial" test utility provided by python-twisted-core fails with the following error message:

Traceback (most recent call last):
File "/usr/bin/trial", line 23, in
from twisted.scripts.trial import run
ImportError: No module named scripts.trial

Investigation reveals that the culprit is an empty "init.py" file generated by the nevow installation. This file is placed in /var/lib/python-support/python2.5/twisted/plugins. The existence of that file breaks twisted's plugin finding approach and causes it to fail to find the twisted.scripts package, which is actually located at /usr/share/pyshared/twisted/scripts.

I can't see how this file gets generated

Source package: nevow
Ubuntu release: Ubuntu 8.04.3 LTS
Package versions: python-nevow: 0.9.26-1, python-twisted-core: 2.5.0-2build2, python2.5: 2.5.2-2ubuntu6

Imported from Launchpad using lp2gh.

• date created: 2009-10-27T03:21:38Z
• owner: zooko
• the launchpad url was https://bugs.launchpad.net/bugs/410098

JSON encoded bytes may include literal (that is, UTF-8 encoded) U+2028 or U+2029. However, JavaScript string literals may not. See http://timelessrepo.com/json-isnt-a-javascript-subset for any more detail that may be interesting.

Athena's JavaScript parser (browser-side JavaScript eval) cannot parse these literals. If we emit escaped versions of the line terminators, the parser won't need to change.

Imported from Launchpad using lp2gh.

• date created: 2011-07-04T23:35:09Z
• owner: exarkun
• the launchpad url was https://bugs.launchpad.net/bugs/805734

There are plenty of json libraries around now. Most of them even support customization for encoding arbitrary objects (such as Nevow's Elements).

Nevow doesn't need to supply its own anymore.

Hi!

I am using Nevow and it's pretty cool. But I encountered a bug with liveglue.js

I am using '#' in hrefs for dynamic contents, so it makes urls like http://host/path#
That is ok, but if user reloads page, liveglue reinitializes and makes all live requests URIs like http://host/path#/livepage_client/... which is completely broken.

Simply added this code to liveglue.js

var anchorIndex = base_url.indexOf('#');
if (anchorIndex != -1) {
base_url = base_url.substring(0, anchorIndex);
}

and everything works.

Imported from Launchpad using lp2gh.

• date created: 2013-06-10T10:19:04Z
• owner: myautneko
• the launchpad url was https://bugs.launchpad.net/bugs/1189412

IE11 identifies itself as

'Mozilla/5.0 (Windows NT 6.3; Trident/7.0; rv:11.0) like Gecko'

which leads subsequently to the wrong assumption of the DivMod UA detection code, that the browser runtime should be set to Firefox with XPath capabilities being present. The error manifests itself by IE11 stating, that XPathResult is undefined.

A possible fix is expanding the test for IE to test for 'Netscape' and no XPathNamespace defined, while the actual 'Netscape' type browser Firefox would then have XPathNamespace defined.

For an explanation of the UA strings used by IE11, please see:
http://blogs.msdn.com/b/ieinternals/archive/2013/09/21/internet-explorer-11-user-agent-string-ua-string-sniffing-compatibility-with-gecko-webkit.aspx

Imported from Launchpad using lp2gh.

• date created: 2013-11-11T05:12:15Z
• owner: werner-thieprojects
• the launchpad url was https://bugs.launchpad.net/bugs/1249946

The tests for LiveFragment spew a lot of deprecation warnings relating to the deprecation of LiveFragment; these should be silenced.

Imported from Launchpad using lp2gh.

• date created: 2011-08-21T03:27:47Z
• owner: mithrandi
• the launchpad url was https://bugs.launchpad.net/bugs/830358

PR #30 changed the log message format in nevow.test.test_appserver.Logging.test_oldStyle to match the format in Twisted 14.0. Unfortunately, this does not match the format used in earlier versions of Twisted, leading to this test failure:

[FAIL]
Traceback (most recent call last):
  File "/«PKGBUILDDIR»/debian/python-nevow/usr/lib/python2.7/site-packages/nevow/test/test_appserver.py", line 193, in test_oldStyle
    ['"fakeaddress2" - - faketime "GET /foo HTTP/1.0" 200 6 '
  File "/usr/lib/python2.7/dist-packages/twisted/trial/_synctest.py", line 356, in assertEqual
    % (msg, pformat(first), pformat(second)))
twisted.trial.unittest.FailTest: not equal:
a = ['fakeaddress2 - - faketime "GET /foo HTTP/1.0" 200 6 "fakerefer" "fakeagent"']
b = ['"fakeaddress2" - - faketime "GET /foo HTTP/1.0" 200 6 "fakerefer" "fakeagent"']


nevow.test.test_appserver.Logging.test_oldStyle

Test coverage is good.

Apart from having moved to Github and the various changes this has entailed (changing how the version number is managed, setting up continuous integration, fixing the test suite against recent Twisted versions), there are also a bunch of fixes made since 0.10 which was almost five years ago. It's time!

This will convince github to render it properly on the project page.

These have been deprecated to follow twisted.web upstream; once they can be removed (the deprecation hasn't been released yet), _DictHeaders should be removed too.

The problem basically is:

gourami:~/.../nevow$ trial nevow.test.test_flatsax.Basic.test_otherns                
nevow.test.test_flatsax
  Basic
    test_otherns ...                                                       [OK]

-------------------------------------------------------------------------------
Ran 1 tests in 0.017s

PASSED (successes=1)


gourami:~/.../nevow$ PYTHONHASHSEED=42 trial nevow.test.test_flatsax.Basic.test_otherns
nevow.test.test_flatsax
  Basic
    test_otherns ...                                                     [FAIL]

===============================================================================
[FAIL]
Traceback (most recent call last):
  File ".../nevow/nevow/test/test_flatsax.py", line 184, in test_otherns
    self.failUnlessEqual(xml, flatten(parseString(xml)))
  File "/usr/lib/python2.7/dist-packages/twisted/trial/_synctest.py", line 356, in assertEqual
    % (msg, pformat(first), pformat(second)))
twisted.trial.unittest.FailTest: not equal:
a = '<html xmlns="http://www.w3.org/1999/xhtml" xmlns:xf="http://www.w3.org/2002/xforms"><p>in default namespace</p><xf:input><xf:label>in another namespace</xf:label></xf:input></html>'
b = '<html xmlns:xf="http://www.w3.org/2002/xforms" xmlns="http://www.w3.org/1999/xhtml"><p>in default namespace</p><xf:input><xf:label>in another namespace</xf:label></xf:input></html>'


nevow.test.test_flatsax.Basic.test_otherns
-------------------------------------------------------------------------------
Ran 1 tests in 0.017s

FAILED (failures=1)

This means that some tests will break some because they depend on some ordering which is not stable . This currently blocks #47 since Tox sets a different hash seed for each run and the probability of getting the "right" seed is probably very low.

I would propose to hunt down the places where such ordering is not guaranteed and use sorted() or something appropriate to fix the problem. (The fix for this test case is too wrap this iteritems() into sorted()). This will make the output more predictable obviously, ease the tests as the tester can expect to have a reliable output, but it has a slight cost performance though.

Other possibilities that I can think of:

• change the tests so they are smarter: instead of doing a basic string equality (or a assert expected in result as I saw in other failing tests where expected content is not in the same order as the one in result), parse the result (they are XML-like strings each times) and assert that the structure is properly formed.
• find the right hash :) (and set it somewhere once and for all, like this for Tox.

Hi,

Nevow is used in the eXeLearning project; I've been working on upgrading it from the ancient version it used before. I think there is a mistake in how the nevow_closeLive function is written:

The function takes an arg by the name of evalAfterDone - which looking at the implementation of nevow_clientToServerEvent is a javascript function that runs after the clientToServerEvent is done, regardless of the result and is not passed to the server.

But in the implementation of nevow_closeLive evalAfterDone is the third parameter, not the second. This results in the evalAfterDone argument being passed to the server as an argument and not treated as evalAfterDone by the nevow_clientToServerEvent function. As far as I can see:

nevow_clientToServerEvent('close', '', evalAfterDone)

Should actually be:

nevow_clientToServerEvent('close', evalAfterDone, '')
or just:
nevow_clientToServerEvent('close', evalAfterDone)

Perhaps I have missed something about how the close event is supposed to be handled... but it seems strange that this argument would be called the same thing in both functions and then switch around.

Will look forward to finding out if this is the case or if I have after reading the docs in livepage misunderstood how close is supposed to work.

Thanks,

-Mike

This is with Nevow 0.10.0 and zope.interface 3.6.4. David-Sarah Hopwood and Drew Perttula each independently reported this error. (I haven't seen it personally, I'm just reporting what they posted.) Drew included a patch to zope.interface work-around it:

http://drewp.quickwitretort.com/2011/07/08/0

http://codepad.org/wd1UUnhh

Imported from Launchpad using lp2gh.

• date created: 2011-07-17T07:02:38Z
• owner: zooko
• the launchpad url was https://bugs.launchpad.net/bugs/811792

It is good when it is easy for folks to be able to read the documentation.

This depends on #20.