Closed Bug 44558 Opened 25 years ago Closed 25 years ago

Regression: Back button does not work at all with blank start page

Categories

(Core :: DOM: Navigation, defect, P1)

x86
Linux
defect

Tracking

()

VERIFIED FIXED

People

(Reporter: 1212mozilla, Assigned: radha)

References

Details

(Keywords: platform-parity, regression, smoketest, Whiteboard: [dogfood+] fix in hand)

In todays nightly build the back button is disabled. It remains greyed out even after visiting multiple sites. This does not appear to be related to frames or any other open bugs involving the back button that I was able to find.
*** Bug 44554 has been marked as a duplicate of this bug. ***
reassigning to history confirming...Alec commented on this on IRC, and there's a dup. I believe it's pp linux...this is a smoketest blocker.
Assignee: asa → radha
Severity: normal → blocker
Status: UNCONFIRMED → NEW
Component: Browser-General → History
Ever confirmed: true
Keywords: pp, regression, smoketest
QA Contact: doronr → claudius
nominating dogfood for obvious reasons...
Keywords: dogfood
I guess windows looks OK. The problem is only in linux? How about Mac? Claudius can you update me? There is no platform specific code in the "SH in frames" feature I enabled over the weekend. My today morning's windows build looks good. Pulling in linux right now
No mac nightlies were made today...and yep, this seems to work fine on win32.
Putting on Putting on [dogfood+] radar.radar.
Whiteboard: [dogfood+]
This works just fine with my Linux commercial build today even on a bugzilla page. Marking WORKSFORME until someone proves me wrong.
Doh! OK, really marking it WORKSFORME this time.
Status: NEW → RESOLVED
Closed: 25 years ago
Resolution: --- → WORKSFORME
welp...alec saw this, tor saw it, the reporter of this bug saw it, and the reporter of the dup saw it...so obviously something happened. Whether or not it's fixed I don't know...or maybe it was never in the comm branch?
Reopening bug. I saw this on my morning build and again in the tree I pulled at about 4pm eastern. Debug build on solaris/native. This debug assertion might from startup might be relevant ("about:blank" is my start page): Adding url about:blank to SH ###!!! ASSERTION: NS_ENSURE_TRUE(NS_SUCCEEDED(AddChildSHEntry(0 , entry, mChildOffset))) failed: '(!((AddChildSHEntry(0 , entry, mChildOffset)) & 0x80000000))', file /cs/src/mozilla/mozilla/docshell/base/nsDocShell.cpp, line 3105 ###!!! Break: at file /cs/src/mozilla/mozilla/docshell/base/nsDocShell.cpp, line 3105
Status: RESOLVED → REOPENED
Resolution: WORKSFORME → ---
rick potts is going to checkin some changes that will take care of the assertion in the debug builds.
Hmmmm, Paul Chen says he had this problem with his mozilla build this morning but after he deleted his ".mozilla" directory everything worked just fine. I, however, didn't have to do this with my commercial build. Weird. Maybe it has something to do with the registry? BTW, someone please make the assert a separate bug and assing it to rpotts. Radha, the PDT wants an ETA in the status whiteboard field of all dogfood+ bugs. Do you have any idea what we should put there? Especially since (I for one) have no clue as to what's causing this yet?
Priority: P3 → P1
Whiteboard: [dogfood+] → [dogfood+] Still investigating, no ETA yet ... stay tuned
Target Milestone: --- → M16
But it looks like Paul Chen's Mac mozilla build from this morning works just fine, like my Linux commercial build. Weird.
Moving my .mozilla directory aside caused the problem to go away.
Since there appears to be a workaround (although unpleasant), can we at least take this off the "blocker" list?
Removing the following lines from prefs.js lets the back button work again: user_pref("browser.startup.homepage", ""); user_pref("browser.startup.page", 0); Hope this helps.
I agree that this is no longer a blocker, especially if the easier workaround that mozilla@casret.com provided works. Lowing severity, if anyone feels otherwise please provide a good reason before bumping it back up.
Severity: blocker → critical
It does appear to be related to using a blank page for your homepage on startup. Removing the lines from your prefs fix the problem. However, if you go into the menus and put blank page back as your startup page, the problem reappears. The summary should be changed to reflect this.
Summary: Regression: Back button does not work at all → Regression: Back button does not work at all with blank start page
This bug affects Mac Mozilla bits from 2000070608. The removing prefs for startpage works as a workaround.
I have a fix for this. Shall checkin when the tree opens.
Whiteboard: [dogfood+] Still investigating, no ETA yet ... stay tuned → [dogfood+] fix in hand
This is related to not adding pages like "about:blank" to SH. Once the fix is checked in, if blank page is your initial page, what you will see is, 1)Start browser (see blank page) 2) Load some page. (The back button will be disabled now because this new page just replaced about:blank in SH. 3)Load some other page. Now back button will come alive and the SH will have entries for page 1 and 2 only, not for about:blank. Some may argue that about:blank s'd also be in SH. I think that's a different issue.
fix checked in.
Status: REOPENED → RESOLVED
Closed: 25 years ago25 years ago
Resolution: --- → FIXED
VERIFIED Fixed it works just as Radha describes in her last comment.
Status: RESOLVED → VERIFIED
Component: History: Session → Document Navigation
QA Contact: claudius → docshell
You need to log in before you can comment on or make changes to this bug.