Closed Bug 39281 Opened 24 years ago Closed 24 years ago

[regression] long links are being truncated (or corrupted)

Categories

(Core :: DOM: HTML Parser, defect, P1)

Product:
Core
Component:
DOM: HTML Parser
Type:
defect

Tracking

()

Status:
VERIFIED FIXED

People

(Reporter: daa, Assigned: rickg)

References

(
URL
)

Details

(Keywords: regression, Whiteboard: Fix checked in...testing)

Attachments

(2 files)

Testcase
270 bytes, text/html
Details
revised testcase
381 bytes, text/html
Details
change to platform/os ALL and P1 fails with 5/15 8AM tip linux and win95 builds
Keywords: dogfood
OS: Linux → All
Priority: P3 → P1
Hardware: PC → All
where shoudl this go? i have a simple testcase <A HREF="http://bugzilla.mozilla.org/bug_status=NEW&d"> http: //bugzilla.mozilla.org/bug_status=NEW&d</A> this has the bug (notice those reversed ???), removing the 'd' after the & fixes it. <A HREF="http://bugzilla.mozilla.org/bug_status=NEW&"> http: //bugzilla.mozilla.org/bug_status=NEW&</A> parser perhaps? setting major and m16.
Severity: blocker → major
Target Milestone: --- → M16
Keywords: regression
ok, that did not work. 2nd trial http://bugzilla.mozilla.org/bug_status=NEW&d has the bug, notice the inverted ??? http://bugzilla.mozilla.org/bug_status=NEW& does not, nothing is trailing the &
Attached file Testcase —
sorry for the spamm, must have been bugzilla parseing that killed my testcases - added an attachment of testcases
somehow, i overwrote sairuh's comments without getting a collision warning - adding them (summary change/cc)
Summary: long links broken → long links are being truncated (or corrupted)
add smoketest keyword
Keywords: smoketest
I cannot view pages on americangreetings.com due to this problem. Also, trying to search anything on google.com gives me an error. I am not sure if it is this problem but when I type something and hit ENTER, then I see that some additional header data is getting appended to the url and an error message is returned. Am passing this bug to event handling for a look since the original problem is for mousing over url's showing weird characters. thnx.
Component: Browser-General → Event Handling
QA Contact: jelwell → janc
reassigning.
Assignee: asadotzler → joki
cc: rickg
Weird things start to happen at the '&' character in URLs. My random guess would be that this bug is related to bug 30386, but I'm grasping at straws here.
If the problem is indeed a "&" in a URI, then this is per the specs. If a URI contains an ampersand (&) then it should be escaped (&amp;) otherwise the parser should interpret it as an entity. For example, in: http://www.domain.com/script?write=yes&copy=no ...the "&copy" bit should be interpreted as "&copy;" which is © (U+00A9). If we do not recognise the entity then "???" is a valid interpretation according to the specs, since they do not specify error handling. So: If this is indeed the problem, then the parser needs to act differently in quirks mode than in strict mode. In quirks mode, it needs to treat unknown entities as an ampersand (i.e. hello&you -> hello&amp;you); and in strict mode, it needs to do something to make authors write "&amp;" -- I would recommend treating &unknown as &#xFFFD; (i.e. hello&you -> hello&#xFFFD;you). U+FFFD is the Unicode REPLACEMENT CHARACTER.
nominating for nsbeta2 is this really event handling?
Keywords: nsbeta2
Ian Hickson - this was working in the nightly before, dunno if that means something. Moving to parser, since the urls are wrong before clicking them (see testcase or my bugs feature on bugzilla). Any idea who this should be assigned to?
Component: Event Handling → Parser
Keywords: smoketest
not to step on anyone's toes - but i'm curious as to why this was bumped down to a severity of major - this make mozilla pretty unusable for browsing any sites (slashdot, deja.com, bugzilla to name a few easy ones) that generate long urls in this format.
Well... I agree that major isn't enough of a designation of this bug's significance, but: Critical crashes, loss of data, severe memory leak Major major loss of function This is a major loss of function.
Keywords: 4xp
rickg, are these related to your DTD changes?
*** Bug 39295 has been marked as a duplicate of this bug. ***
I'll take a look to see if this is me. I sure hope not. :)
Assignee: joki → rickg
Fix in hand.
Status: NEW → ASSIGNED
rickg, please attach -or checkin- the fix. This blocks the PSM smoketests.
Severity: major → blocker
Blocks: 39295
It looks like the fix was just checked in (via bonzai), waiting for status update.
Whiteboard: Fix checked in...testing
*** Bug 39451 has been marked as a duplicate of this bug. ***
Landed fix. Sorry for the error to begin with.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
This is not completely fixed. Reopening. It works for incorrect HTML, but it's broken for correct HTML (that uses &amp;). Any occurrence of &amp; shows up as in the URL as an & but cuts off the rest of the URL. See the revised testcase above.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
*** Bug 39456 has been marked as a duplicate of this bug. ***
Nice catch david; fix in hand -- will land shortly.
Fixed.
Status: REOPENED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
*** Bug 39501 has been marked as a duplicate of this bug. ***
*** Bug 39504 has been marked as a duplicate of this bug. ***
*** Bug 39550 has been marked as a duplicate of this bug. ***
*** Bug 39588 has been marked as a duplicate of this bug. ***
*** Bug 39583 has been marked as a duplicate of this bug. ***
*** Bug 39605 has been marked as a duplicate of this bug. ***
*** Bug 39636 has been marked as a duplicate of this bug. ***
verified fixed on win98 200051708 mozilla build
Status: RESOLVED → VERIFIED
Reopening - I'm getting this with win32 2000052908. It is somewhat different..I simply get a | instead of the long string, including .cgi. However, the url loads correctly in mozilla, just the urlbar shows the corrupted url (this, however, makes reload impossible, as it uses the text in the urlbar and does not find the page). Lowering severity to MAJOR, as this is not a blocker and not critical (not a crasher)
Severity: blocker → major
Status: VERIFIED → REOPENED
Keywords: regression
Resolution: FIXED → ---
Summary: long links are being truncated (or corrupted) → [regression] long links are being truncated (or corrupted)
btw - is this still Parser, or should this be moved?
sorry, this is totally unrelated. Going to open a newbug. Sorry for the spam!
Severity: major → blocker
Status: REOPENED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
returning to verified fixed since reopening was a mistake.
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: