35300 2010-02-23 09:33:02 -0800 (GTK) HTTP 307 after a 303 after a POST re-sends POST data from the original request 2017-10-21 19:22:23 -0700 1 1 1 Unclassified WebKit Page Loading 528+ (Nightly build) All All RESOLVED WONTFIX P2 Normal --- 1 beidson webkit-unassigned berto commit-queue danw gustavo mcatanzaro mrobinson webkit.review.bot xan.lopez youennf zan oldest_to_newest 192986 0 beidson 2010-02-23 09:33:02 -0800 HTTP 307 after a 303 after a POST re-sends POST data from the original request. See https://bugs.webkit.org/show_bug.cgi?id=31410 for more discussion and the fix that went in for Mac and Windows. GTK needs to do the same. 789712 1 berto 2012-12-12 11:04:29 -0800 Looks like in GTK the second redirection doesn't even work (I get a timeout). I'm taking a look at this. 805262 2 danw 2013-01-11 07:30:35 -0800 So with the changes from bug 88961, we now make the correct series of requests. However, the test still fails because DRT doesn't print the right strings on redirect. Eg, -expected has: http://127.0.0.1:8000/loading/resources/post-to-303-target.php - willSendRequest <NSURLRequest URL http://127.0.0.1:8000/loading/resources/303-to-307-target.php, main document URL http://127.0.0.1:8000/loading/resources/303-to-307-target.php, http method GET> redirectResponse <NSURLResponse http://127.0.0.1:8000/loading/resources/post-to-303-target.php, http status code 303> but we output: http://127.0.0.1:8000/loading/resources/303-to-307-target.php - willSendRequest <NSURLRequest URL http://127.0.0.1:8000/loading/resources/303-to-307-target.php, main document URL http://127.0.0.1:8000/loading/resources/303-to-307-target.php, http method GET> redirectResponse <NSURLResponse http://127.0.0.1:8000/loading/resources/post-to-303-target.php, http status code 303> (The first URL is wrong.) 828554 3 youennf 2013-02-08 07:27:44 -0800 EFL has a similar bug (https://bugs.webkit.org/show_bug.cgi?id=93214) for which I submitted a patch. As the code is shared with Gtk (libsoup backend), this patch may be of interest. I actually ran DRT GTK with and without the patch. Without the patch, the last request is a POST (possible regression?). With the patch, the last request is a GET, as expected. As of the DRT URL bug, the WebKitWebResource private uri field is changed in FrameLoaderClient::dispatchWillSendRequest. I was wondering whether it would be a more appropriate model to keep this field unchanged accross redirections? If so, with a small change to willSendRequestCallback, this bug would probably be fixed. I can provide a patch along those lines. 870065 4 196679 berto 2013-04-05 14:05:06 -0700 Created attachment 196679 Patch (In reply to comment #3) > As of the DRT URL bug, the WebKitWebResource private uri field is > changed in FrameLoaderClient::dispatchWillSendRequest. > I was wondering whether it would be a more appropriate model to keep > this field unchanged accross redirections? Yes, this actually seems to be the problem. Do we need to change that field? The identifier is supposed to be associated to the initial request. It works fine if we do it like that, here's the patch, I tried all http/tests and it doesn't seem to break anything. 870070 5 webkit.review.bot 2013-04-05 14:08:27 -0700 Attachment 196679 did not pass style-queue: Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'LayoutTests/ChangeLog', u'LayoutTests/platform/gtk-wk1/TestExpectations', u'Source/WebKit/gtk/ChangeLog', u'Source/WebKit/gtk/WebCoreSupport/FrameLoaderClientGtk.cpp', u'Tools/ChangeLog', u'Tools/DumpRenderTree/gtk/DumpRenderTree.cpp']" exit_code: 1 Tools/DumpRenderTree/gtk/DumpRenderTree.cpp:1284: Weird number of spaces at line-start. Are you using a 4-space indent? [whitespace/indent] [3] Total errors found: 1 in 4 files If any of these errors are false positives, please file a bug against check-webkit-style. 871903 6 berto 2013-04-09 09:25:38 -0700 One other possibility if we don't want to change the current behavior is to add a new 'original-uri' property to the WebKitWebResource class. I can easily prepare a patch. (In reply to comment #4) > Do we need to change that field? The identifier is supposed to be > associated to the initial request. One thing to note, though, is that we're currently updating the uri property (which is read-only) by directly accessing the private structure from a different class... Gustavo wrote the original code (bug 29134), adding him to Cc. 873418 7 commit-queue 2013-04-10 18:32:18 -0700 Attachment 196679 did not pass style-queue: Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'LayoutTests/ChangeLog', u'LayoutTests/platform/gtk-wk1/TestExpectations', u'Source/WebKit/gtk/ChangeLog', u'Source/WebKit/gtk/WebCoreSupport/FrameLoaderClientGtk.cpp', u'Tools/ChangeLog', u'Tools/DumpRenderTree/gtk/DumpRenderTree.cpp']" exit_code: 1 Tools/DumpRenderTree/gtk/DumpRenderTree.cpp:1284: Weird number of spaces at line-start. Are you using a 4-space indent? [whitespace/indent] [3] Total errors found: 1 in 4 files If any of these errors are false positives, please file a bug against check-webkit-style. 873588 8 youennf 2013-04-11 00:27:42 -0700 (In reply to comment #6) > One other possibility if we don't want to change the current behavior > is to add a new 'original-uri' property to the WebKitWebResource > class. Does that new property be useful outside this test? If created for the purpose of the test, I would avoid doing that. A temporary workaround may be to create an expected file for that test. We would at least check that there is no regression in the POST-303-307 behavior. > One thing to note, though, is that we're currently updating the uri > property (which is read-only) by directly accessing the private > structure from a different class... Yes, when reading that code, I thought it might be better to encapsulate this as a function call, if this code is to remain. 873784 9 196679 gustavo 2013-04-11 06:39:54 -0700 Comment on attachment 196679 Patch Yeah, the URI is to remain read-only for API users, it should only be updated by internal code, that is why it's done the way it is. I'm OK with adding an internal call to update it instead of poking the private structure, though. I think adding a new property for original-uri is a good idea, since it maintains the API expectations while still providing complete information. All that said, wk1 is in maintenance mode and there was a decision not to add new API to it unless strictly necessary, so I wonder, do we really want to fix this test at this point? r- for the change in API expectation, adding a new property sounds like a better solution. 873792 10 berto 2013-04-11 06:52:55 -0700 (In reply to comment #9) > All that said, wk1 is in maintenance mode and there was a decision > not to add new API to it unless strictly necessary, so I wonder, do > we really want to fix this test at this point? I think you're right, actually. 1363114 11 mcatanzaro 2017-10-21 19:22:23 -0700 Closing since this looks like an old WK1 bug. 196679 2013-04-05 14:05:06 -0700 2013-04-11 06:39:54 -0700 Patch 35300.diff text/plain 4669 berto ZGlmZiAtLWdpdCBhL0xheW91dFRlc3RzL0NoYW5nZUxvZyBiL0xheW91dFRlc3RzL0NoYW5nZUxv ZwppbmRleCBhOWYwYTI5Li41OGQzMmU5IDEwMDY0NAotLS0gYS9MYXlvdXRUZXN0cy9DaGFuZ2VM b2cKKysrIGIvTGF5b3V0VGVzdHMvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTQgQEAKKzIwMTMtMDQt MDUgIEFsYmVydG8gR2FyY2lhICA8YWdhcmNpYUBpZ2FsaWEuY29tPgorCisgICAgICAgIChHVEsp IEhUVFAgMzA3IGFmdGVyIGEgMzAzIGFmdGVyIGEgUE9TVCByZS1zZW5kcyBQT1NUIGRhdGEgZnJv bSB0aGUgb3JpZ2luYWwgcmVxdWVzdAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9z aG93X2J1Zy5jZ2k/aWQ9MzUzMDAKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMh KS4KKworICAgICAgICBSZW1vdmUgZXhwZWN0YXRpb24gZm9yIDMwNy1hZnRlci0zMDMtYWZ0ZXIt cG9zdCwgaXQgd29ya3MgZmluZSBub3cuCisKKyAgICAgICAgKiBwbGF0Zm9ybS9ndGstd2sxL1Rl c3RFeHBlY3RhdGlvbnM6CisKIDIwMTMtMDQtMDUgIFphbiBEb2JlcnNlayAgPHpkb2JlcnNla0Bp Z2FsaWEuY29tPgogCiAgICAgICAgIFVucmV2aWV3ZWQgR1RLIGdhcmRlbmluZy4KZGlmZiAtLWdp dCBhL0xheW91dFRlc3RzL3BsYXRmb3JtL2d0ay13azEvVGVzdEV4cGVjdGF0aW9ucyBiL0xheW91 dFRlc3RzL3BsYXRmb3JtL2d0ay13azEvVGVzdEV4cGVjdGF0aW9ucwppbmRleCBkZmNmOTdiLi45 MTFmNWIyIDEwMDY0NAotLS0gYS9MYXlvdXRUZXN0cy9wbGF0Zm9ybS9ndGstd2sxL1Rlc3RFeHBl Y3RhdGlvbnMKKysrIGIvTGF5b3V0VGVzdHMvcGxhdGZvcm0vZ3RrLXdrMS9UZXN0RXhwZWN0YXRp b25zCkBAIC0yMTUsOSArMjE1LDYgQEAgQnVnKEdUSykgZmFzdC9sb2FkZXIvcmVsb2FkLXBvbGlj eS1kZWxlZ2F0ZS5odG1sIFsgRmFpbHVyZSBdCiBCdWcoR1RLKSBmYXN0L2xvYWRlci9zdG9wLXBy b3Zpc2lvbmFsLWxvYWRzLmh0bWwgWyBGYWlsdXJlIF0KIEJ1ZyhHVEspIGh0dHAvdGVzdHMvc2Vj dXJpdHkvZmVlZC11cmxzLWZyb20tcmVtb3RlLmh0bWwgWyBUaW1lb3V0IF0KIAotIyBOZWVkcyB0 byBtYWtlIHN1cmUgdGhlIHJlZGlyZWN0LWNoYWluIHNjZW5hcmlvIGluIGh0dHBzOi8vYnVncy53 ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD0zMTQxMCB3b3Jrcwotd2Via2l0Lm9yZy9iLzM1MzAw IGh0dHAvdGVzdHMvbG9hZGluZy8zMDctYWZ0ZXItMzAzLWFmdGVyLXBvc3QuaHRtbCBbIEZhaWx1 cmUgXQotCiAjIE1pc3NpbmcgRFJUIEFQSS4KIEJ1ZyhHVEspIGh0dHAvdGVzdHMvbG9hZGluZy9w cmVsb2FkLXNsb3ctbG9hZGluZy5waHAgWyBGYWlsdXJlIF0KIApkaWZmIC0tZ2l0IGEvU291cmNl L1dlYktpdC9ndGsvQ2hhbmdlTG9nIGIvU291cmNlL1dlYktpdC9ndGsvQ2hhbmdlTG9nCmluZGV4 IDlkZGY5ZDYuLmVkMWFkNWUgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJLaXQvZ3RrL0NoYW5nZUxv ZworKysgYi9Tb3VyY2UvV2ViS2l0L2d0ay9DaGFuZ2VMb2cKQEAgLTEsMyArMSwxNiBAQAorMjAx My0wNC0wNSAgQWxiZXJ0byBHYXJjaWEgIDxhZ2FyY2lhQGlnYWxpYS5jb20+CisKKyAgICAgICAg KEdUSykgSFRUUCAzMDcgYWZ0ZXIgYSAzMDMgYWZ0ZXIgYSBQT1NUIHJlLXNlbmRzIFBPU1QgZGF0 YSBmcm9tIHRoZSBvcmlnaW5hbCByZXF1ZXN0CisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQu b3JnL3Nob3dfYnVnLmNnaT9pZD0zNTMwMAorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgIERvbid0IGNoYW5nZSB0aGUgcmVzb3VyY2UgdGhhdCBjb250YWlu cyB0aGUgVVJJIG9mIHRoZSBpbml0aWFsCisgICAgICAgIHJlcXVlc3QgaWYgdGhlcmUncyBhIHJl ZGlyZWN0aW9uLgorCisgICAgICAgICogV2ViQ29yZVN1cHBvcnQvRnJhbWVMb2FkZXJDbGllbnRH dGsuY3BwOgorICAgICAgICAoV2ViS2l0OjpGcmFtZUxvYWRlckNsaWVudDo6ZGlzcGF0Y2hXaWxs U2VuZFJlcXVlc3QpOgorCiAyMDEzLTA0LTA0ICBNYXJ0aW4gUm9iaW5zb24gIDxtcm9iaW5zb25A aWdhbGlhLmNvbT4KIAogICAgICAgICBbR1RLXSBSZW1vdmUgdGhlIGd5cCBidWlsZApkaWZmIC0t Z2l0IGEvU291cmNlL1dlYktpdC9ndGsvV2ViQ29yZVN1cHBvcnQvRnJhbWVMb2FkZXJDbGllbnRH dGsuY3BwIGIvU291cmNlL1dlYktpdC9ndGsvV2ViQ29yZVN1cHBvcnQvRnJhbWVMb2FkZXJDbGll bnRHdGsuY3BwCmluZGV4IDkzMTg4OGEuLjg5NDJhYTIgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9XZWJL aXQvZ3RrL1dlYkNvcmVTdXBwb3J0L0ZyYW1lTG9hZGVyQ2xpZW50R3RrLmNwcAorKysgYi9Tb3Vy Y2UvV2ViS2l0L2d0ay9XZWJDb3JlU3VwcG9ydC9GcmFtZUxvYWRlckNsaWVudEd0ay5jcHAKQEAg LTI0NiwxMyArMjQ2LDYgQEAgdm9pZCBGcmFtZUxvYWRlckNsaWVudDo6ZGlzcGF0Y2hXaWxsU2Vu ZFJlcXVlc3QoV2ViQ29yZTo6RG9jdW1lbnRMb2FkZXIqIGxvYWRlciwKICAgICBXZWJLaXRXZWJS ZXNvdXJjZSogd2ViUmVzb3VyY2UgPSB3ZWJraXRfd2ViX3ZpZXdfZ2V0X3Jlc291cmNlKHdlYlZp ZXcsIGlkZW50aWZpZXJTdHJpbmcuZ2V0KCkpOwogICAgIEdSZWZQdHI8V2ViS2l0TmV0d29ya1Jl cXVlc3Q+IG5ldHdvcmtSZXF1ZXN0KGFkb3B0R1JlZihraXROZXcocmVxdWVzdCkpKTsKIAotICAg IGlmICghcmVkaXJlY3RSZXNwb25zZS5pc051bGwoKSkgewotICAgICAgICAvLyBUaGlzIGlzIGEg cmVkaXJlY3QsIHNvIHdlIG5lZWQgdG8gdXBkYXRlIHRoZSBXZWJSZXNvdXJjZSdzIGtub3dsZWRn ZQotICAgICAgICAvLyBvZiB0aGUgVVJJLgotICAgICAgICBnX2ZyZWUod2ViUmVzb3VyY2UtPnBy aXYtPnVyaSk7Ci0gICAgICAgIHdlYlJlc291cmNlLT5wcml2LT51cmkgPSBnX3N0cmR1cChyZXF1 ZXN0LnVybCgpLnN0cmluZygpLnV0ZjgoKS5kYXRhKCkpOwotICAgIH0KLQogICAgIGdfc2lnbmFs X2VtaXRfYnlfbmFtZSh3ZWJWaWV3LCAicmVzb3VyY2UtcmVxdWVzdC1zdGFydGluZyIsIG1fZnJh bWUsIHdlYlJlc291cmNlLCBuZXR3b3JrUmVxdWVzdC5nZXQoKSwgbmV0d29ya1Jlc3BvbnNlLmdl dCgpKTsKICAgICBnX3NpZ25hbF9lbWl0X2J5X25hbWUobV9mcmFtZSwgInJlc291cmNlLXJlcXVl c3Qtc3RhcnRpbmciLCB3ZWJSZXNvdXJjZSwgbmV0d29ya1JlcXVlc3QuZ2V0KCksIG5ldHdvcmtS ZXNwb25zZS5nZXQoKSk7CiAKZGlmZiAtLWdpdCBhL1Rvb2xzL0NoYW5nZUxvZyBiL1Rvb2xzL0No YW5nZUxvZwppbmRleCBjYWFjMzY5Li43ODgzZTAxIDEwMDY0NAotLS0gYS9Ub29scy9DaGFuZ2VM b2cKKysrIGIvVG9vbHMvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTYgQEAKKzIwMTMtMDQtMDUgIEFs YmVydG8gR2FyY2lhICA8YWdhcmNpYUBpZ2FsaWEuY29tPgorCisgICAgICAgIChHVEspIEhUVFAg MzA3IGFmdGVyIGEgMzAzIGFmdGVyIGEgUE9TVCByZS1zZW5kcyBQT1NUIGRhdGEgZnJvbSB0aGUg b3JpZ2luYWwgcmVxdWVzdAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1 Zy5jZ2k/aWQ9MzUzMDAKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKwor ICAgICAgICBHZXQgdGhlIHJlcXVlc3QgVVJJIGZyb20gdGhlIHJlc291cmNlIHBhcmFtZXRlciwg d2hpY2ggbm93CisgICAgICAgIGNvbnRhaW5zIHRoZSBVUkkgb2YgdGhlIGluaXRpYWwgcmVxdWVz dCBpZiB0aGVyZSdzIGEgcmVkaXJlY3Rpb24uCisKKyAgICAgICAgKiBEdW1wUmVuZGVyVHJlZS9n dGsvRHVtcFJlbmRlclRyZWUuY3BwOgorICAgICAgICAod2lsbFNlbmRSZXF1ZXN0Q2FsbGJhY2sp OgorCiAyMDEzLTA0LTA0ICBLZW50IFRhbXVyYSAgPHRrZW50QGNocm9taXVtLm9yZz4KIAogICAg ICAgICBSZW1vdmUgQ2hyb21pdW1QdWJsaWNBcGkgcGF0dGVybiwgYW5kIGNjL21lc3NhZ2UgcnVs ZXMgZm9yIGl0LgpkaWZmIC0tZ2l0IGEvVG9vbHMvRHVtcFJlbmRlclRyZWUvZ3RrL0R1bXBSZW5k ZXJUcmVlLmNwcCBiL1Rvb2xzL0R1bXBSZW5kZXJUcmVlL2d0ay9EdW1wUmVuZGVyVHJlZS5jcHAK aW5kZXggY2VlNDYxYy4uYjNlZjVlYSAxMDA2NDQKLS0tIGEvVG9vbHMvRHVtcFJlbmRlclRyZWUv Z3RrL0R1bXBSZW5kZXJUcmVlLmNwcAorKysgYi9Ub29scy9EdW1wUmVuZGVyVHJlZS9ndGsvRHVt cFJlbmRlclRyZWUuY3BwCkBAIC0xMjgxLDcgKzEyODEsNyBAQCBzdGF0aWMgdm9pZCB3aWxsU2Vu ZFJlcXVlc3RDYWxsYmFjayhXZWJLaXRXZWJWaWV3KiB3ZWJWaWV3LCBXZWJLaXRXZWJGcmFtZSog d2ViRgogCiAgICAgaWYgKCFkb25lICYmIGdUZXN0UnVubmVyLT5kdW1wUmVzb3VyY2VMb2FkQ2Fs bGJhY2tzKCkpCiAgICAgICAgIHByaW50ZigiJXMgLSB3aWxsU2VuZFJlcXVlc3QgJXMgcmVkaXJl Y3RSZXNwb25zZSAlc1xuIiwKLSAgICAgICAgICAgICAgIGNvbnZlcnROZXR3b3JrUmVxdWVzdFRv VVJMUGF0aChyZXF1ZXN0KS5kYXRhKCksCisgICAgICAgICAgICAgICBjb252ZXJ0V2ViUmVzb3Vy Y2VUb1VSTFBhdGgocmVzb3VyY2UpLmRhdGEoKSwKICAgICAgICAgICAgICAgIGRlc2NyaXB0aW9u U3VpdGFibGVGb3JUZXN0UmVzdWx0KHJlcXVlc3QpLmRhdGEoKSwKICAgICAgICAgICAgICAgIGRl c2NyaXB0aW9uU3VpdGFibGVGb3JUZXN0UmVzdWx0KHJlc3BvbnNlKS5kYXRhKCkpOwogCg==