I've been thinking about stream converters, content-encoding, and the
interaction of necko with consumers for a bit now... and I'm somewhat at a
loss for ideas, so I thought I'd float what I have so far and see what people
think. So:
Goal: The necko consumer should havecontrol over the
1. Can I use aWindowTarget to display the results in the correct
window? Would I do a QueryInterface on it to get some kind of window
object? If so, which interface would I use?
Are you creating a DOM tree and then trying to attach it to a window? I don't
think that's possible yet...
2.
I've been thinking about a few issues we have with content-type detection:
1) Our internal detection is not too reliable (for file://, ftp://, etc).
2) There is often context at channel open that could provide a better guess at
the type (eg the type attribute of link, a, or script).
At the
var nsIURI Components.classes[
@mozilla.org/network/standard-url;1
].createInstance(Components.interfaces.nsIURI);
nsIURI.spec = file:///tmp/test.file;
This is the wrong way to create a URI and in particular it creates a file://
uri that does not know that it's a file uri (does not QI to
Chris Sobotkiewicz wrote:
However, sometimes things go nuts and
a single user can cause my server to grind to a halt. What do I do?
I'd hate to deny Netscape users access to the play button, but its
getting to that point!
So your server grinds to a halt because of requests to see whether the
Darin Fisher wrote:
Another option would be to have necko automagically wedge an instance of
nsUnknownDecoder between the file channel and its listener. This is
already done for HTTP (and elsewhere i think). BZ might even have a bug
filed about doing this.
Not really. One interesting related
Also, is there any context in which you want to know the file type but not
actually use it?
Yes. For example, when implementing the accept attribute of input
type=file.
More to the point, there are many cases where you need to know the file type in
one part of the code and actually use the
I was worried this would appear next week. :-) To me, the accept attribute
seems to be designed by someone who thinks that a name defines an object,
which is silly.
Where do names appear? It's designed by someone who assumes that given an
object you can get meta-information about its type
I hope you're just being theoretical. Do you really want mozilla to open every
file it doesn't understand just to see what's in it? And what do you do with
a mime type you don't know about?
This whole discussion is theoretical so far. The question is what theory we
should select and put into
So I decided to write a short document about it, and put it up at
http://stud4.tuwien.ac.at/~e0225227/MozillaMIMETypes.html
Comments welcome. Hopefully, it will be useful to someone.
First, thanks a ton for doing this.
Second, we should consider putting this on mozilla.org and linking to
(changed the things I did not commetn on. document still at
http://stud4.tuwien.ac.at/~e0225227/MozillaMIMETypes.html)
One issue with the first paragraph -- it uses content type and mime [sic]
type to mean the same thing. I would use MIME type both places.
The target audience is mostly web
OK, how about I put this document at:
http://www.mozilla.org/docs/web-developer/mimetypes.html
And link to it from
http://www.mozilla.org/docs/web-developer/ (in the Writing for Mozilla
section)
and
http://www.mozilla.org/projects/netlib/ (the core documentation section)
and maybe
It's not a matter of the wrong mime type, it's a matter of the wrong file.
Agreed. That should indeed be addessed. Not very easy to do, though...
I wouldn't object to associated types as long as there's a distinction between a
stream of unknown type which is treated as a default type and a
Let's assume that mozilla uses application/octet-stream for any file it
doesn't understand. That's the default type. I'm saying that treating a file
(or any object) as application/octet-stream is different from knowing the
file is application/octet-stream and mozilla should keep these cases
However, even after scouring the header files, I can't figure out how
nsICacheEntryDescriptor structure is formatted.
That's not a structure; that's an interface defined in
http://lxr.mozilla.org/seamonkey/source/netwerk/cache/public/nsICacheEntryDescriptor.idl
Hope that helps,
Boris
--
A
Christian Biesinger wrote:
That requires code to get from a .url file to the URL it links to.
I assume that's not covered by the existing symlink stuff on nsIFile?
string instead of nsIURI because this is xpcom (= no necko), and
because the current callers want a string.
So. this is my
[EMAIL PROTECTED] wrote:
Yeah, this is a good theory, but as far as I can tell 1.2.1 doesn't have
enough frozen interfaces to do async socket IO reading and writing
with only frozen interfaces...
That's unfortunately correct.
And unless I am missing something, it looks like there are STILL not
Raj Saini wrote:
1. I write a implentation of the nsIURIContentListener and register with
nsIURILoader during xomponent registration.
Or even just register with the category manager... see
http://www.mozilla.org/docs/docshell/uri-load-start.html#CategoryManager
2. Use the DoContent method of
Raj Saini wrote:
I have written a component. I am looking for some notification where I
can get the conent type of a request.
For what request? It's not clear to me what exactly you are trying to
do
-Boris
___
Mozilla-netlib mailing list
[EMAIL
Raj Saini wrote:
For the second part that is diabling some of UI
operations, what I want is open a new window when the
contents types are text/x-html-encrypted arrives. my
converter should decrypt the contents and pump them to
the new browser window for rendering.
The problem is that there is
Andrew Van Uitert wrote:
I need to have my filter sit at a level where it sees all http requests
before they hit the socket. The reason I have to sit so low, is that many of
the porn sites use re-directs to pay other sites for their listings. At the
XPCOM embed level I can only see the original
Doug Turner wrote:
What is the bug number?
Bug 191839
-Boris
___
Mozilla-netlib mailing list
[EMAIL PROTECTED]
http://mail.mozilla.org/listinfo/mozilla-netlib
Christian Biesinger wrote:
How about this:
* codefalse/code if the URI Loader
should
* continue handling the load and call the
returned streamlistener's methods.
Yeah, looks good. r=me (sr too, if you want).
-Boris
Matthew Wilson wrote:
And then the converter creates a new nsIChannel with the correct
content type, and passes that to the listener which the converter
recieved during AsyncConvertData - is that right?
Well, it doesn't necessarily create a new channel. It may
SetContentType on the old channel
Nanjun Hu wrote:
Can I use nsIWebBrowserPersist to do such save as?
That's what the browser UI uses... See
http://lxr.mozilla.org/seamonkey/source/xpfe/communicator/resources/content/contentAreaUtils.js#691
then call persist-SaveDocument?
Yep.
I tried to do it in the above, but it fails. So any
lasco wrote:
Could you give me an example how to make a valid path passing to
InitWithPath? suppose I want to save the page as b.html and pics in b_files.
InitWithPath takes an absolute path.
-Boris
___
Mozilla-netlib mailing list
[EMAIL PROTECTED]
But will using saveURI get the HTML text from server again or using the
loaded copy, since the webpage is already loaded?
Depends on the cache flags you set on the persistence object.
Boris
--
Washington [D.C.] is a city of Southern efficiency and
Northern charm.
--
I searched in the source tree, but I am afraid that I didn't find the caller
of saveDocument and saveURL. So would you...? thanks!!!
I'm afraid, no. Please make sure that you searched the .xul and .js files.
Boris
--
Washington [D.C.] is a city of Southern efficiency and
Northern charm.
Nanjun Hu wrote:
After calling nsIWebBrowser::SaveDocument method
There is no such method on nsIWebBrowser.
-Boris
___
Mozilla-netlib mailing list
[EMAIL PROTECTED]
http://mail.mozilla.org/listinfo/mozilla-netlib
maybe you can use the address of the nsIChannel object?
And get notified on redirects, right?
Boris
--
Bradley's Bromide:
If computers get too powerful, we can organize them
into a committee -- that will do them in.
___
Mozilla-netlib mailing
It seems that mozilla asks for files in parallel (which is ok), but until at least
start of next file arrives, it does not read other streams either (which
is questionable, imho).
Script file:
document.write('boo');
HTML file:
html
body
Hey script src=boo.js!
/body
/html
Note that we can't
What I didnt' find by looking in the Mozilla's network code is a way to
uniquely identify a request and be able to find its parent.
necko does not know what the parent is, in general (and it's not really
well-defined in many cases anyway). We need to somehow fix that, but that's a
huge
Yes, but I would like to avoid that. If page has 2 identical pictures,
then I want onModifyRequest/onExamineResponse calls for every of them,
regardles of their source (network, local cache).
With the image cache, what actually happens is that the load request is never
even sent to necko. The
Phil Hühn wrote:
Can I get some help? I'm doing file (web page) processing - I create an
object which needs to validate the page URL (as is done for 'save page
as'). So this object implements component:
@mozilla.org/network/urichecker;1
So your object registers itself for this contract,
Phil Hühn wrote:
So your object registers itself for this contract, overriding the
Mozilla urichecker?
Yes, it does register itself for this contract 'override the Moz
urichecker'?? Is there one already (that I can use)?
Well, what does calling createInstance() on that contractid give you
Bryan White wrote:
I have put togehter a stripped down version here:
http://pg2.arcamax.com/cgi-bin/redirtest
Thanks. This is a bug in the XULDocument code in Mozilla,
unfortunately. See http://bugzilla.mozilla.org/show_bug.cgi?id=240358.
You may be able to work around it in the interim by
Bryan White wrote:
You may be able to work around it in the interim by using a
client-side redirect (meta refresh); I _think_ that won't set the
originalURI to anything untoward, unlike this case.
Could you explain this a little, I don't know what you mean here.
What happens is that on HTTP
Phil Hühn wrote:
I'm actually modifying the code in contentAreaUtils.js - when I started
this message thread I thought I'd try to leave out stuff like that to
simplify the question but maybe I made it worse ;-)
Heh. ;)
So the 'this' value is (should be!) a different nsHeaderSniffer object
in
Bryan White wrote:
Sorry the location should be:
jar:https://secure.arcamax.com/moztest/moztest2.jar!/test.xul
OK, there I get (in the JS console):
Signature Verification Error: the signature on test.xul is invalid
because the digital signature (*.RSA) file is not a valid signature of
the
Phil Hühn wrote:
Two dialogs pop up (i.e one for each tab)
Mm... that could be uncool. Which OS is this? You're testing on?
Or do you mean that even if you suppress the filepickers things still
don't work?
-Boris
___
Mozilla-netlib mailing list
Phil Hühn wrote:
Yes, I suppress the file pickers and things don't work.
OK. Is this in Seamonkey or Firefox? Or both?
In either case, file a bug please; it sounds like things are racing for
access to some sort of shared resource, with bad side-effects. Attach
whatever diffs are needed to
Bryan White wrote:
Sorry the location should be:
jar:https://secure.arcamax.com/moztest/moztest2.jar!/test.xul
Bryan, could you put the enablePrivelege() call back? I thought I'd
broken something with a patch to fix one of the problems you raised,
since it wasn't coming up, but it looks like
Darin Fisher wrote:
Call, Read and check what it returns. If it returns NS_OK and zero
bytes read, then that means EOF.
We should really document that in the IDL
-Boris
___
Mozilla-netlib mailing list
[EMAIL PROTECTED]
Brett Goodman wrote:
Can you tell me if Mozilla uses pluggable protocols
It does.
and if so, is it similar to IE.
I don't know what IE's looks like, so...
Can you point me to some docs that describe how to implement one?
http://www.mozilla.org/projects/netlib/new-handler.html
is a start.
Brett Goodman wrote:
Does it matter that this PPH will only be for the Windows platform, and that
the code will be proprietary?
I'm not a lawyer, but as far as I understand the answer is no.
So I wouldn't be checking in any code, just distributing the DLL.
Right. All that really means is that
Olivier Cahagne wrote:
After fix for bug 109553 was committed, one major server failed because
of HTTP content-location badly configured: bug 231072.
Content-location support has been backed out due to the number of servers that
break with it and the fact that it breaks anchor traversals in
David Herron wrote:
But in .../ipc/src/nsIPipeTransport.h it has this code which bombs the compile
#ifdef MOZILLA_VERSION
public nsIInputStreamNotify,
public nsIOutputStreamNotify
#else
// Mods for Mozilla version prior to 1.3b
Andre Pang wrote:
What would be the best way to go about
doing this? (Sorry, I'm a bit new to Mozilla development practices.)
Should I submit a bug report now, or use the mailing lists (such as this
one) to discuss possible solutions first before submitting one?
Either one works fine, really.
naltrexone wrote:
Oh, one other observation-- I think the mapping *does* work for files
on the local machine. Just not for files on remote servers.
The extension to type mapping is only used when there is no MIME type provided
for the data. The servers are almost certainly providing the data
Bjorn Thor Jonsson wrote:
file = files.getNext()
.QueryInterface(Components.interfaces.nsILocalFile);
jsFile = new File(file.path); // JSLib file
jsFile.open();
Er... why not just create an input stream here?
var sData = clientfile= + encodeURIComponent( jsFile.read()
Christian Biesinger wrote:
That won't work, apparently - XMLHTTPRequest overwrites the notification
callbacks you set.
I filed bug 292391 on making it do so in a less destructive way.
Neil, could maybe you test the patch I posted and let me know whether it
resolves your problem?
-Boris
Neil Stansbury wrote:
Interestingly enough though, it doesn't work... or rather it does but
the events fire all in one go after the request is finished! It's
almost like the thread the events are on is blocking.
Are you doing a sync or async XMLHttpRequest?
-Boris
LeRoyK wrote:
I am working on Javascript instrumentation for Firefox 1.0x using a
chrome extension.
How are you registering your nsIWebProgressListener?
I have never recorded the STATUS_SENDING_TO event. Does that one still
occur?
The socket transport still sends it, yes
-Boris
LeRoyK wrote:
I do not see the nsIWebProgressListener interface used at all
when the loadGroup is null.
Right. The docloader is hooked up to channels through the loadgroup.
I really don't know what the problem you're running into is... :(
-Boris
___
LeRoyK wrote:
Would the next logical step be to create a simple test case and submit
it as a bug, or are there other options I might try first?
If you can create a simple testcase that shows the problem, that sounds like the
best next step, yes.
-Boris
LeRoyK wrote:
This change would slow down every nsHttpTransaction:Init() call by the
time it takes to check a preference.
Actually, if you cached the pref value (on nsHttpHandler, probably, and made
sure the pref is observed by the handler by adding things in nsHttpHandler::Init
as needed),
Ken Theng wrote:
Hi all ,
issit possible to load this nsIDOMDocument into the
nsiWebNavigation ? Cos i need to render the nIDOMDocument i gotten
I don't believe you can stick a DOMDocument into a given nsIWebNavigation at the
moment
-Boris
Ken wrote:
Do you know of anyway to render the DOMDocument i gotten ?
There isn't one, really. You can move its documentElement into another document
(the one in the webnavigation), but that's not guaranteed to work right,
especially in 1.7 and earlier builds.
-Boris
Andrea wrote:
I think that firefox does not support multipart http responses. Is it
correct?
What makes you think that? We have code to support multipart responses, and it
works fine last I checked (for example, bugzilla query results are multipart
responses).
-Boris
Andrea wrote:
Hi,
I took a look at the bugzilla query results, which is a multipart
response and works fine.
But, the type of multipart support that I was talking about is
rendering of images included in the http response with multipart.
You mean that the image and the HTML are all part of the
Michael Vincent van Rantwijk wrote:
It's available at any time. If you're asking at what point it could
depend on what the server returned, that's after OnStartRequest or
after open() returns.
I hoped that, but that's not the case because it throws JS errors on the
JS console if you try to
Michael Vincent van Rantwijk wrote:
I don't think that there's anything guaranteed to work in this huge and
constantly changing code base, but at least it works right now.
The things that are guaranteed to work are documented as such with frozen
interfaces or frozen contracts.
-Boris
Michael Vincent van Rantwijk wrote:
Yikes, I assumed to be using some random fake domain name
There are standard fake domain names to use, as documented in the DNS RFCs (eg
example.com) if you need a fake domain name.
-Boris
___
Mozilla-netlib
Tomaz Nolêto wrote:
When mozilla starts dealing with a not renderizable file using
nsExternalHelperAppService, I need to get file size info after the
Show() method is called from a nsIHelperAppLauncherDialog. Can I do it
inside Show() itself? What are the ways to do it?
First problem is that
zen wrote:
I writting a simple XPCOM with the following code but the server script
cannot detect the postdata. Could anyone please help? Is this a bug
since the loadURI is actually doing the POST but the server cannot
detect any postdata.
Does Gecko send the data to the server? Can you
zen wrote:
FYI, loadURI() construct the postdata (*refer code from 1st posting) as
HTTP data while form action=POST construct as Line-based text data
(both monitored from ethereal).
I'm not sure what that means. Since you have a \r\n at the beginning of your
string, the data should end up
zen wrote:
Yes, the data end up in the header, i think it not consider as the
content of that particular HTTP post. Please refer screenshot attached.
loadURI_post.gif - capture during loadURI performing post
form_post.gif - capture during form action=post submited
The main difference I
I had figured out that channels are a single shot data connection. So
I need to handle the concept of being connected in my channel.
Yes, channels are single-shot things.
SOAP doesn't appear to use channels, or stream listeners so I'm some
what forcing a square peg into round hole.
I think
Neil Stansbury wrote:
Oh ok, I couldn't find a SOAP method that gave me access to those
interfaces. I'll look again.
There might well not be one
I'm still trying to understand your setup. You're implementing a
protocol handler, right?
Yep an XML message/rpc protocol, using (usually)
zen wrote:
is there any alternative method that similar to loadURI which can accept
postdata as argument and is a FROZEN interface?
No.
-Boris
___
Mozilla-netlib mailing list
Mozilla-netlib@mozilla.org
http://mail.mozilla.org/listinfo/mozilla-netlib
Neil Stansbury wrote:
I don't want to have to implement multiple Message.Delete() methods for
each underlying XML format. I wanted to transfer a generic request of
Delete() to the channel for that connection, and then have it handle
the Delete() specifics for that connection type.
This
71 matches
Mail list logo