Closed
Bug 41241
Opened 24 years ago
Closed 24 years ago
<form target=_foo method=post> sends POST and GET request
Categories
(Core :: DOM: Core & HTML, defect, P3)
Tracking
()
VERIFIED
FIXED
mozilla0.9
People
(Reporter: naujok, Assigned: adamlock)
References
()
Details
(Whiteboard: [nsbeta3-][rtm-])
Attachments
(5 files)
144 bytes,
text/html
|
Details | |
12 bytes,
text/plain
|
Details | |
3.87 KB,
patch
|
Details | Diff | Splinter Review | |
3.54 KB,
patch
|
Details | Diff | Splinter Review | |
2.61 KB,
patch
|
Details | Diff | Splinter Review |
I am building an internal project which includes an HTTP server. Certain commands are then run internally within the server by get or post handlers. One of the forms (full HTML to follow) is sent using a POST method. However, Mozilla (M15) sends it using the GET method (Full HTTP headers included). This only happens on certain forms within the project (The log-in screen works correctly, the HTML for that follows as well.) In any case, calling GET instead of PUT is obviously incorrect. (The project is called NIWS 3.0 if you're wondering.) HTML of Log-In Screen [greeting.html] (Works) ------------------------------------------------------------------------------ <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1"> <META NAME="Author" CONTENT="Jeffrey R. Naujok"> <META NAME="GENERATOR" CONTENT="EditPlus 1.25"> <META NAME="Description" CONTENT="Home page for NIWS 3.0"> <link rel=stylesheet type="text/css" href="/style-sheets/NIWS3Standard.css" title="NIWS3Standard"> <TITLE>Welcome to NIWS 3.0!</TITLE> </HEAD> <BODY bgcolor="white" onLoad="document.LoginForm.UserName.focus();"> <table width=600 border=0 cellpadding=0 cellspacing=0> <tr> <td align=left width=200 class=loginMenu valign=top> <b>NIWS 3.0</b><br> <hr width=80% height=2 color=#E0E0E0> Login to NIWS<br> <a href="/newuser/index.html">Get a NIWS Account</a><br> <hr width=80% height=2 color=#E0E0E0> <a href="/intro/whatis.html">What is NIWS?</a><br> <a href="/intro/whatsnew.html">What's new in 3.0?</a><br> <hr width=80% height=2 color=#E0E0E0> <a href="http://niwsroom.mcit.com">The NIWSROOM</a><br> <hr width=80% height=2 color=#E0E0E0> <a href="/documents/support.html">Support</a><br> <a href="/documents/comments.html">Comments</a><br> <hr width=80% height=2 color=#E0E0E0> <a href="/documents/browsers.html">Supported Browsers</a><br> <hr width=80% height=2 color=#E0E0E0> <a href="/documents/legal_information.html">Legal Information</a><br> <br><br> <div class=infoText> Thanks for coming to the NIWS system!<br> NIWS has been developed entirely within the MCI WorldCom corporation using no outside vendor products.<br> </div> </td> <td align=center valign=center> <h1>Welcome to the NIWS 3.0 System</h1> <br> <hr width=80% color=#E0E0E0> <h2>Please Log-in</h2> <TABLE BORDER=0 CELLSPACING=0 CELLPADDING=0> <TR> <TD ALIGN=CENTER NOWRAP WIDTH="400"> <TABLE BORDER=2 CELLSPACING=0 CELLPADDING=0 WIDTH="100%" class=loginFrame> <TR> <TD ALIGN=CENTER WIDTH="100%" class=loginFrame> <FORM action="exec/LoginTest" method="POST" name="LoginForm" target="_top"> <br> <TABLE BORDER=0 CELLSPACING=0 CELLPADDING=0 WIDTH="100%" class=loginFrame> <TR> <TD ALIGN=RIGHT WIDTH="300" class=loginFrame> <font face=arial> <nobr> User ID: <INPUT type="text" size="15" name="UserName"> </nobr> <br> <nobr> PassWord: <INPUT type="password" size="15" name="PassWord"> </nobr> </font> </TD> <TD ALIGN=CENTER WIDTH="125" class=loginFrame> <INPUT type="submit" name="Login" value="Login"> <br><br> <INPUT type="reset" name=" Clear" value="Clear"> </TD> </TR> </TABLE> </FORM> </TD> </TR> </TABLE> </TD> </tr> </TABLE> <br><a href="/forgotpassword/index.html">Forget your password?</a><br> </td> </tr> </table> <table width=600 border=0 cellpadding=0 cellspacing=0> <tr> <td class=copyrightBar align=center valign=center> Copyright © 2000, MCI WorldCom Corporation. All Rights Reserved. </td> </tr> </table> </BODY> </HTML> HTML of Circuit Long-Route Form [index.html] (Does not work) -------------------------------------------------------------------------------- <!-- Data was parsed by NIWS 3.0 --> <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1"> <META NAME="Author" CONTENT="Jeffrey R. Naujok"> <META NAME="GENERATOR" CONTENT="EditPlus 1.25"> <META NAME="Description" CONTENT="Circuit Long Route Report"> <link rel=stylesheet type="text/css" href="/style-sheets/NIWS3Standard.css" title="NIWS3Standard"> <title>NIWS: Circuit Long-Route Report</title> <script language="javascri
Comment 2•24 years ago
|
||
reporter: can you please ensure that this problem is still occurring on the latest nightly builds? you mentioned that you saw it in M15, which is old by now
Reporter | ||
Comment 3•24 years ago
|
||
With last night's build (6/4/2000), I'm still getting the same error. Please note that this only happens if the target of the form is "_blank". If I use "_top", or a named frame, it doesn't happen. [GET /exec/DoReport HTTP/1.1] [Host: stutz.mcit.com:8080] [User-Agent: Mozilla/5.0 (Windows; U; WinNT4.0; en-US; m16) Gecko/20000604] [Accept: */*] [Accept-Language: en] [Accept-Encoding: gzip,deflate,compress,identity] [Keep-Alive: 300] [Connection: keep-alive] []
Updated•24 years ago
|
Status: NEW → ASSIGNED
Summary: Form POST is sent with HTTP GET method instead of POST → <form target=_blank method=post> sends GET request
Comment 6•24 years ago
|
||
Comment 7•24 years ago
|
||
This behaviour will happen regardless of the target, as long as it is not an existing window/frame, and thus causes a new one to open. We actually send out *two* requests to the server. The first is correctly a POST, fired from: nsWebShell::HandleLinkClickEvent() The response seems to be dropped on the floor in a messy way, as we get: ###!!! ASSERTION: OnDataAvailable fired after OnStopRequest: 'mOnStopFired == PR _FALSE', file d:\builds\pollmann\mozilla\netwerk\protocol\http\src\nsHTTPRespons eListener.cpp, line 1227 ###!!! Break: at file d:\builds\pollmann\mozilla\netwerk\protocol\http\src\nsHTT PResponseListener.cpp, line 1227 The second is a GET request with no data, encoded or posted. This is sent from: GlobalWindowImpl::OpenInternal where we call newDocShell->LoadURI(uriToLoad, loadInfo); My best guess is that we are also sending two requests for <a target="_foo">bar</a> too because it uses the same code path. Nobody probably notices because they would both be GET's Since the problem seems to be in GlobalWindowImpl and DocShell, giving this one to you Adam. If I can help, please let me know!
Assignee: pollmann → adamlock
Status: ASSIGNED → NEW
Summary: <form target=_blank method=post> sends GET request → <form target=_foo method=post> sends POST and GET request
Docshell will only use "POST" if there is some postdata to post. Do any of these forms have no data? Regarding "PUT" - I'm rather worried that I can't find any reference to this at all! I have a horrible feeling that docshell has not support for it at all. Does anyone know any different?
Comment 10•24 years ago
|
||
What am I smoking? There's no such protocol as "PUT" - just "GET" AND "POST".
Comment 11•24 years ago
|
||
heh...I dunno what you're smoking but it's either really powerful or yer takin' a lot of it. (I also noticed that your testcase only seems to consist of the words "Hello World"...was that intentional?)
Comment 12•24 years ago
|
||
Adam, the problem is not with POST in general, or forms, but a POST request that has a target set for a window that does not exist yet. The problem is that two requests are sent. 1) Correctly a post, sends all the right data. 2) When the new window opens to display the results, it sends a second request. This second request is a GET. This is the bad one.
Assignee | ||
Comment 13•24 years ago
|
||
Yes the test case was intentional :) I was using the submit attachment CGI to test Mozilla's POST behaviour.
Comment 14•24 years ago
|
||
minusing. although this is serious. the embedding plate is full.
Whiteboard: [nsbeta3-]
Reporter | ||
Comment 15•24 years ago
|
||
My only problem with you marking this as a "future concern" is that the product I'm working on where I originally found this problem represents over 1000 users at MCI WorldCom. They're already itching to find a reason to make MicroDolts their official browser, and since we open a new blank window to provide reporting results /every time/, this will give them the excuse to drop Netscape entirely from MCI WorldCom. That's 50,000 users. Still seem like this should be shelved?
Comment 16•24 years ago
|
||
*** Bug 51259 has been marked as a duplicate of this bug. ***
Comment 17•24 years ago
|
||
*** Bug 51259 has been marked as a duplicate of this bug. ***
Comment 20•24 years ago
|
||
*** Bug 49494 has been marked as a duplicate of this bug. ***
Comment 21•24 years ago
|
||
*** Bug 53019 has been marked as a duplicate of this bug. ***
Comment 22•24 years ago
|
||
Changing OS to All. This bug is crashing linux build 2000-10-03-09-MN6 and crashing Mac build 2000-09-29-11-MN6.
OS: Windows NT → All
Comment 23•24 years ago
|
||
rtm-, outside the specific environment where this was discovered it seems very unlikely to be hit. If you are working on a patch, please update the bug.
Whiteboard: [nsbeta3-] → [nsbeta3-][rtm-]
Comment 24•24 years ago
|
||
*** Bug 59431 has been marked as a duplicate of this bug. ***
Comment 25•24 years ago
|
||
*** Bug 60388 has been marked as a duplicate of this bug. ***
Comment 26•24 years ago
|
||
cc'ing self... selmer@netscape.com: with all the dupes, I would think it would be safe to say that this bug will be found in other areas outside the reporters development environment... I use target="new" to open a new form for the users, posting the information that they have entered into my screen. This bug is a pain to work around. My current hack/workaround is to open a new HTML form using window.open. In my case, the elements to be submitted are always the same, so I have hard coded them in this new form. I set these element's values via Javascript and then do a javascript form.submit(). This seems to work well, although future maintainers of this code will wonder what drugs I was on. In more complicated circumstances, doing some form of innerHTML tweaking could prolly get this workaround to be more robust... <RANT type="frustration"> In my opinion, this is just another bug that makes it hard to develop for Mozilla. Netscape needs to realize that without developers actively supporting their platform, they'll continue to tank in market share, especially in corporate environments. Making it hard for developers to do simple things will continue to drive them away which will speed up the spiral... </RANT>
Comment 27•24 years ago
|
||
Is bug 59547 a dupe of this bug? If so, there are a lot of instances this bug is causing pain and confusion.
Comment 28•24 years ago
|
||
*** Bug 59547 has been marked as a duplicate of this bug. ***
Comment 29•24 years ago
|
||
To clarify, it looks like 59547 is a dup, but the other instances are probably duplicates of bug 27006.
Assignee | ||
Comment 30•24 years ago
|
||
It looks to me like the channel loading and window creation behaviour is a bit screwy. This stack shows where the newly created "_blank" target window is told to load the URI: nsDocShell::DoURILoad(nsDocShell * const 0x05e26300, nsIURI * 0x05a41720, nsIURI * 0x00000000, nsISupports * 0x00000000, int 0x00000000, int 0x00000000, const char * 0x00000000, nsIInputStream * 0x00000000, nsIInputStream * 0x00000000) line 3095 nsDocShell::InternalLoad(nsDocShell * const 0x05e26300, nsIURI * 0x05a41720, nsIURI * 0x00000000, nsISupports * 0x00000000, int 0x00000000, int 0x00000000, const char * 0x00000000, nsIInputStream * 0x00000000, nsIInputStream * 0x00000000, unsigned int 0x00000001, nsISHEntry * 0x00000000) line 3005 + 47 bytes nsDocShell::LoadURI(nsDocShell * const 0x05e26300, nsIURI * 0x05a41720, nsIDocShellLoadInfo * 0x00000000, unsigned int 0x00000000) line 429 + 65 bytes GlobalWindowImpl::OpenInternal(GlobalWindowImpl * const 0x0455b590, JSContext * 0x0455b370, long * 0x010b0f48, unsigned int 0x00000002, int 0x00000000, nsIDOMWindowInternal * * 0x0012f6d8) line 3132 GlobalWindowImpl::Open(GlobalWindowImpl * const 0x0455b594, JSContext * 0x0455b370, long * 0x010b0f48, unsigned int 0x00000002, nsIDOMWindowInternal * * 0x0012f6d8) line 2067 nsBrowserContentHandler::HandleContent(nsBrowserContentHandler * const 0x05a418f0, const char * 0x05a41930, const char * 0x01e9f348, const char * 0x05a39740, nsISupports * 0x045530b0, nsIChannel * 0x05a39a60) line 1857 + 84 bytes nsURILoader::DispatchContent(nsURILoader * const 0x017a8e80, const char * 0x05a41930, int 0x00000007, const char * 0x05a39740, nsIChannel * 0x05a39a60, nsISupports * 0x00000000, nsIURIContentListener * 0x04553030, nsISupports * 0x045530b0, char * * 0x0012f96c, nsIURIContentListener * * 0x0012f974, int * 0x0012f964) line 1027 + 53 bytes nsDocumentOpenInfo::DispatchContent(nsIChannel * 0x05a39a60, nsISupports * 0x00000000) line 310 + 165 bytes nsDocumentOpenInfo::OnStartRequest(nsDocumentOpenInfo * const 0x05a39780, nsIChannel * 0x05a39a60, nsISupports * 0x00000000) line 241 + 16 bytes nsHTTPFinalListener::OnStartRequest(nsHTTPFinalListener * const 0x05a39620, nsIChannel * 0x05a39a60, nsISupports * 0x00000000) line 1131 nsHTTPServerListener::FinishedResponseHeaders() line 1056 + 48 bytes nsHTTPServerListener::OnDataAvailable(nsHTTPServerListener * const 0x05a37f50, nsIChannel * 0x05a3fd54, nsISupports * 0x05a39a60, nsIInputStream * 0x05a370c0, unsigned int 0x00000000, unsigned int 0x00000000) line 427 + 8 bytes nsOnDataAvailableEvent::HandleEvent(nsOnDataAvailableEvent * const 0x05a407f0) line 407 + 47 bytes nsStreamListenerEvent::HandlePLEvent(PLEvent * 0x05a40800) line 92 + 12 bytes PL_HandleEvent(PLEvent * 0x05a40800) line 576 + 10 bytes PL_ProcessPendingEvents(PLEventQueue * 0x00abfb10) line 509 + 9 bytes _md_EventReceiverProc(HWND__ * 0x00140c90, unsigned int 0x0000c115, unsigned int 0x00000000, long 0x00abfb10) line 1054 + 9 bytes USER32! 77e71820() 00abfb10() The interesting stuff starts to happen in nsBrowserContentHandler::HandleContent. This is what gets called when data starts to arrive. This method pops a few things such as the URL and the window target into a Javascript context and tells the global window manager to open the window. It then aborts the original channel!? Next, the global window manager creates the new window and docshell (because it doesn't exist) and tells the docshell to load the URI provided in the JS context when HandleContent called to open the window. So there are a couple of weird things happening here: 1. The original POST operation is aborted. Why? Can a channel to be bound to a newly created docshell? 2. The global window methods Open & OpenInternal are not passed channel or postdata information so they open the new window as a GET request. If it's possible to attach the original channel to a docshell then that should be done. The open methods on the global window should have channel arguments and the browser content handler should pass that over rather than aborting the original binding. This stops two URL requests being submitted including the erroneous second GET operation.
Assignee | ||
Comment 31•24 years ago
|
||
Assignee | ||
Comment 32•24 years ago
|
||
The attachment solves the "create one POST channel, open a new window when content arrives, create a GET channel and abort the POST channel" issue. It does this by creating the new window much earlier. It creates the new window just as the URI loader object is about to open the first POST channel while it's finding the window context it should be associated with. The new code creates the new window, sets the window context and changes the target to itself so the channel content goes straight into the new window and doesn't hit the old code. It works but I'm not familiar with the nuances of script contexts to know if I may have broken something else, such as when I create a new window with an empty string URI in the JS context. Is that safe? I'm also unsure about the dependencies I've introduced. Now the uriloader project depends on the JS library and drags some DOM window stuff in too. Is that safe? Should this "create a new window" code I've introduced really be a service in the DOM library? Comments please.
Comment 33•24 years ago
|
||
Looks like this is probably a good thing to me. CC'ing Rick Potts though, because he has more knowledge of URI loading...
Comment 34•24 years ago
|
||
Browser: Mozilla M18, dated 05 January 2001. (It was the last build for the moment). Reports itself as: Mozilla/5.0 (Windows; U; Win98; en-US; m18) Gecko/20010104 I found the similar problem when requesting a HTML document via the POST when the server requires authorization (note no "target" attribute is used!). The original HTML and the exact HTTP headers follow: ------------------------------------------------ [The file enter.htm]: <html> <head> <TITLE>APCIS2000: Entrance</TITLE></head> <body BGCOLOR="#000066" BACKGROUND="/apcis/imgs/bkgrMainMenu.gif"> <TABLE><TR><TD width=750> <nobr> <table BORDER="0" WIDTH="100%"> <tr> <td COLSPAN="3"align="center"><font SIZE="-2" FACE="Arial" COLOR="#FFCCCC"> The Entrance page </font> </td> </tr> <tr> <td><img SRC="/apcis/imgs/Globe.gif" WIDTH="113" HEIGHT="105" ALIGN="bottom"></td> <td></td> <td><p ALIGN="right"><font SIZE="+2" COLOR="#000066"><b><i>APCIS2000 </i></b></font></td> </tr> <tr> <td COLSPAN="3"><p align="center"><font SIZE="+2" COLOR="#FF9999"> Asia - Pacific<br> Computerized Information System<br> on Port State Control<BR> </font></td> </tr> </table> <NOSCRIPT> <table BORDER="0" BGCOLOR="#000066" CELLSPACING="12" CELLPADDING="10" WIDTH="100%"> <tr><td align="center"><b><font SIZE="+2" COLOR="#ED4E56"> Warning! JavaScript is disabled.<br> <i>APCIS2000</i> requires JavaScript for its operations.<br> Please, <a HREF="jsenabling.htm"><font COLOR="#9999FF">enable JavaScript</font></a> in your browser and reload this page. </font></b></td></tr> </table> </NOSCRIPT> <form ACTION="/apcis/FMPro" METHOD="post" name="RegForm"> <input type="hidden" name="-DB" value="Access.fp3"> <input type="hidden" name="-Lay" value="web"> <input type="hidden" name="-Format" value="preload.htm"> <input type="hidden" name="-Error" value="oops.htm"> <input type="hidden" name="UserIP" value="192.168.0.101"> <input type="hidden" name="UserName" value="testUser"> <input type="hidden" name="Key" value="33469064753"> <input type="hidden" name="-New" value=""> </form> <DIV ALIGN="CENTER"><BR><BR> <A HREF="javascript:document.RegForm.submit()" >ENTER</A> <BR><BR></DIV> <table BORDER="0" BGCOLOR="#000066" CELLSPACING="0" CELLPADDING="2" WIDTH="100%" BACKGROUND="/apcis/imgs/tblbkgrMainMenu.gif"> <tr> <td> <p ALIGN="right"><b> <a HREF="mailto:support@apcis.tmou.org"><font SIZE="-1" FACE="Arial" COLOR="#000066">mail to tech support</font></a><font SIZE="-1" FACE="Arial" COLOR="#000066"> </font></b> </td> </tr> </table> <BR> <table BORDER="0" ALIGN="CENTER"> <tr> <td ALIGN="center"><font COLOR="#FF9999" FACE="Arial">Optimized for screen resolution = 800x600</font></td> </tr> </table> </nobr> </TD></TR></TABLE> </body> </html> -------------------------- [Exact HTTP headers]: {First Request}: POST /apcis/FMPro HTTP/1.1 Referer: http://127.0.0.1:8000/apcis/FMPro?-db=access.fp3&-lay=web&-format=enter.htm&-tok en=FrVTqX126zCwHsz&-view Host: 127.0.0.1:8000 User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; m18) Gecko/20010104 Accept: */* Accept-Language: ru,en Accept-Encoding: gzip,deflate,compress,identity Keep-Alive: 300 Connection: keep-alive Content-type: application/x-www-form-urlencoded Content-Length: 129 -DB=Access.fp3&-Lay=web&-Format=preload.htm&-Error=oops.htm&UserIP=%25UserIP%25& UserName=%25UserName%25&Key=FrVTqX126zCwHsz&-New= {Response} HTTP/1.0 401 Server: FileMakerPro/4.0 WWW-Authenticate: Basic realm="Database access.fp3" Content-type: text/html {Second request} GET /apcis/FMPro HTTP/1.1 Authorization: Basic dGVzdHVzZXI6dGVzdA== Host: 127.0.0.1:8000 User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; m18) Gecko/20010104 Accept: */* Accept-Language: ru,en Accept-Encoding: gzip,deflate,compress,identity Keep-Alive: 300 Connection: keep-alive {Response} HTTP/1.0 200 OK Server: FileMakerPro/4.0 MIME-Version: 1.0 Content-type: text/html <HTML><HEAD> <TITLE>Unexpected Error</TITLE> <BODY> <H1>Unexpected Error: No arguments passed!</H1> </BODY> </HTML> ------------------------------------------------- Note that both Netscape 4.xx, and MSIE 5 have no such a problem.
Comment 35•24 years ago
|
||
Hmm I'm nervous about adding uriloader dependencies we add with this patch. Reading Adam's comments I see he's nervous about that too. Can anyone thing of any other way we could get the same behavior without adding these dependencies? I can't think of anything off the top of my head.
Comment 36•24 years ago
|
||
I forgot to cc myself after I made my comments.
Assignee | ||
Comment 38•24 years ago
|
||
Comment 39•24 years ago
|
||
The code itself looks good to me. Do you need to change any Mac or unix projects now that the uriloader needs to link with JS? Also, about the hidden window part (and the appshell service), is this part necessary? I'm trying to think if we could get in a state where we are running a url with a _new or _blank window target that didn't originate from another window.....
Comment 40•24 years ago
|
||
*** Bug 67433 has been marked as a duplicate of this bug. ***
Assignee | ||
Comment 41•24 years ago
|
||
The mac project will need updating, Unix should be fine since it links to js already. I've taken out the hidden window fallback code and it seems to function correctly. I can't think of a way that a "_blank" target window could be opened without a context so this is probably a good removal. Scott, please sr the new attachment which follows.
Assignee | ||
Comment 42•24 years ago
|
||
Comment 43•24 years ago
|
||
since the hidden window code has been taken out, you should be able to remove: +#include "nsIAppShellService.h" +#include "nsAppShellCIDs.h" + +static NS_DEFINE_CID(kAppShellServiceCID, NS_APPSHELL_SERVICE_CID); sr=mscott if you take those lines out...
Assignee | ||
Comment 44•24 years ago
|
||
Thanks Scott, fix is checked in now
Status: ASSIGNED → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
Comment 45•24 years ago
|
||
Verifying on builds Windows 98: 2001-02-06-08-Mtrunk Linux RedHat6.2: 2001-02-06-08-Mtrunk MacOS 8.x: 2001-02-05-10-Mtrunk
Status: RESOLVED → VERIFIED
Comment 46•24 years ago
|
||
Is this really fixed in the latest nightlies? I see in some sites Mozilla is sending two requests when document.FOO.submit() is used to submit a form as in this example below. <html> <head> <meta http-equiv="content-type" content="text/html;charset=x-sjis"> <title></title> </head> <a href="javaScript:document.yourForm.submit();">submit</a></p> <form name="yourForm" action="http://www2.france.co.jp/france/kensaku.asp" method="post"> <input type="hidden" name="SHINKAN" value="1"><input type="hidden" name="DISPMODE" value="TEXT"> </form> </body> </html>
Comment 47•24 years ago
|
||
The above example is now WFM in 2001020908 build.
Comment 48•24 years ago
|
||
Browser: Mozilla 0.8, dated 12 February 2001. (It was the last build for the moment). Reports itself as: Mozilla/5.0 (Windows; U; Win98; en-US; 0.8) Gecko/20010212 The problem has not been fixed. I've got the same result (HTTP headers follow): [Request1] POST /apcis/FMPro HTTP/1.1 Referer: http://127.0.0.1:8000/apcis/FMPro?-db=Access.fp3&-lay=web&-format=enter.htm&-tok en=tst&-view Host: 127.0.0.1:8000 User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; 0.8) Gecko/20010212 Accept: */* Accept-Language: ru, en Accept-Encoding: gzip,deflate,compress,identity Keep-Alive: 300 Connection: keep-alive Content-type: application/x-www-form-urlencoded Content-Length: 117 -DB=Access.fp3&-Lay=web&-Format=preload.htm&-Error=oops.htm&UserIP=%25UserIP%25& UserName=%25UserName%25&Key=tst&-New= [Response1] HTTP/1.0 401 Unauthorized Server: FileMaker Pro/4.0 WWW-Authenticate: Basic realm="Protected resource" [Request2] GET /apcis/FMPro HTTP/1.1 //POST SHOULD BE HERE INSTEAD OF GET!!!!!!! Authorization: Basic b3dsLXRzdDpvd2w= Host: 127.0.0.1:8000 User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; 0.8) Gecko/20010212 Accept: */* Accept-Language: ru, en Accept-Encoding: gzip,deflate,compress,identity Keep-Alive: 300 Connection: keep-alive [Response2] HTTP/1.0 200 OK MIME-Version: 1.0 Server: APCIS Entrance Gateway/1.2 Content-Type: text/html <HTML> <HEAD> <TITLE>Unexpected Error</TITLE> </HEAD> <BODY> Unexpected Error: No arguments passed! </BODY> </HTML> I found the problem when requesting a HTML document via the POST when the server requires authorization (note no "target" attribute is used!) Thanks.
Status: VERIFIED → REOPENED
Resolution: FIXED → ---
Assignee | ||
Comment 49•24 years ago
|
||
The original problem which was to do with target windows is fixed so I'm remarking it as such. I have opened bug 68666 to deal with the Javascript submit problems.
Status: REOPENED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
Comment 51•24 years ago
|
||
*** Bug 65982 has been marked as a duplicate of this bug. ***
Comment 52•24 years ago
|
||
This only fixes the problem for target="_new" and target="_blank". The problem remains for target="newwindow" when there is no existing window named 'newwindow'. See bug 72491.
Comment 53•24 years ago
|
||
I'm experiencing something like this when submitting forms to a PerlEx script (similar to mod_perl, but for Windows, from activestate.com). I'll submit a form to a very simple PerlEx script that simply prints out the variables that were passed to the script, as well as the %ENV variables. It shows as a GET, and no data coming from the browser. Then I hit the back button, submit the form again, and it works correctly this time, showing as a POST and all the form variables come through OK. I am using a recent version of CGI.pm to parse the form input. I have duplicated this bug on: - Mozilla 0.8 (Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; 0.8) Gecko/20010215) - Netscape 6.0.1 (where I first got complaints of this happening) - Mozilla Nightly Build as of 3/30/2001 (Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; 0.8.1)) This is causing us serious problems. We are going to have to recommend that our users not use Netscape 6 if this doesn't get fixed soon. (Of course, it would be nice if we could point them to a new build of Mozilla which works, as a replacement, instead of MSIE.)
Status: VERIFIED → REOPENED
Resolution: FIXED → ---
Updated•24 years ago
|
Target Milestone: mozilla0.8 → mozilla0.9
Assignee | ||
Comment 54•24 years ago
|
||
I'm closing this bug in favour of bug 72491. This bug dealt with problems with _blank and _new which were fixed. The other bug deals with problems using GET/POST to named windows. Please CC yourself to the other bug.
Status: REOPENED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
Comment 55•24 years ago
|
||
verifying fixed on build 2001-04-10-04-trunk Windows 98
Status: RESOLVED → VERIFIED
Updated•5 years ago
|
Component: HTML: Form Submission → DOM: Core & HTML
You need to log in
before you can comment on or make changes to this bug.
Description
•