Start Firefox in Safe Mode {web link} While you are in safe mode;

Type about:preferences#advanced<Enter> in the address bar.

Under Advanced, Select General. Look for and turn off Use Hardware Acceleration.

Poke around safe web sites. Are there any problems?

>>Type about:preferences#advanced<Enter> in the address bar.<

Setting that made no difference. I ran 30-40 test searches on Google with that set and Firefox restarted in Safe Mode with all add-ons disabled. The problem persists on the vast majority of searches. However backing up to Google search results after following some links doesn't result in the problem.

For instance when I do a Google search from www.google.com (USA) on "Steve Gibson" and click the first result, which for me is his Wikipedia page: "Steve Gibson (computer programmer) - Wikipedia". If I hit CTRL/+ a couple times to zoom text after the page loads, and then click Firefox's 'Back" icon, the previous Google search result page will be empty.

However if I click on the 2nd result in Google's search results, which for me is "GRC | Steve Gibson's WebZone:", I can zoom the text several times and then click back to the Google search results and find them all displayed fine.

But I've been having this problem when backing up to Google search results probably 70-90% of the time since I updated from Firefox v33 to v34.

It's frustrating because Firefox seems to remember the text zoom level for entire domains. So if I zoom on one Wikipedia page, Firefox will display all Wikipedia pages I visit subsequently at that zoom level. I used to love this feature. But it's become a real pain now when I back up to Google.

BTW: If in the 1st example above, I first reset the text zoom level while I'm still on the Steve Gibson Wiki page, all the Google search results are displayed fine after I click Firefox's 'Back" icon. I don't have this problem in Chrome or Opera.

Sorry, this seems to be an issue with Firefox's in-memory cache: when you use a high zoom level on Google with "instant prediction results" (the default setting) then Firefox glitches in redisplaying the page after going back (you may see the results for a split second before they blank out).

Assuming you like your zoom level and like using instant prediction results, then the workaround is to turn off part of Firefox's in-memory caching by changing a setting under the covers.

(1) In a new tab, type or paste about:config in the address bar and press Enter. Click the button promising to be careful.

(2) In the search box above the list, type or paste sess and pause while the list is filtered

(3) Double-click the browser.sessionhistory.max_total_viewers preference and change the value from -1 to 0 (that's a zero).

(Please don't change any other similar-sounding settings to 0 or you may disable Firefox's ability to save/restore your open windows and tabs.)

Then reload the Google results page and test.

Fixed? (Or should I say, worked-around?)

Incidentally, the default is to store and apply your zoom level by site, so changing the zoom level on Wikipedia would have no effect on Google, or vice versa. In the old days, Firefox applied zoom level changes on a tab-by-tab basis. I wonder whether you somehow switched back to that system? You can check here:

(1) In a new tab, type or paste about:config in the address bar and press Enter. Click the button promising to be careful.

(2) In the search box above the list, type or paste zoom and pause while the list is filtered

(3) If the browser.zoom.siteSpecific preference is bolded and "user set" to false, then Firefox is using the old method. If you prefer to set the level by site, right-click > Reset this setting.

If you use any zoom-related extensions, those might affect this independently, but I'm not aware of any Firefox 34-specific issues (not that it's possible for me or anyone to read all the threads here...).

Thanks for your input jscher2000.

>>this seems to be an issue with Firefox's in-memory cache: when you use a high zoom level on Google with "instant prediction results" (the default setting)<<

Yes.. The Google homepage does seem to default to "instant prediction results", but I haven't set Firefox to a higher level of zoom for the Google search page.

>>Double-click the browser.sessionhistory.max_total_viewers preference and change the value from -1 to 0 (that's a zero). <<

Ah... That workaround does seem to work for me. I hope the developers fix this for future releases though. As I mentioned, I hadn't seen this problem before the release of v34.

>>If the browser.zoom.siteSpecific preference is bolded and "user set" to false, then Firefox is using the old method.<<

No... That's still set to the default value of 'true'. I'm up to date on that one.

Thanks again for your help sorting this out for me jscher2000. It was really driving me nuts. Pass the word on to the devs if you have any inside track to them.

Hi mozfellow, there is an old report in the bug tracking system for this, but I'm not sure anyone has figured out how Google's instant prediction layout triggers this bug (as you can imagine, Google's scripts are complex and difficult to read).

jscher2000 said

there is an old report in the bug tracking system for this..

How old? Pre v34?

but I'm not sure anyone has figured out how Google's instant prediction layout triggers this bug..

Not only complex, but probably like much of technology these days, esp. being Google, changing at a rate that makes it hard to stay on top of. For some reason I seem to have missed the route to accessing this forum's blockquote & editable HTML on my Win7 system... 8-O ... :D

Hi mozfellow, you asked how old the bug is: the report dates back to Firefox 23 and was filed on 8/22/2013. (908446)

That's quite a while back. Can't help but wonder how changes between 33 and 34 brought the problem to the surface for me. Something related to the complexities of both Google's code and FF's code as you pointed out I guess.

In writing computer code, sometimes everything goes Great ! ! No problems. Other times, you need to twitch it a bit. And other times . . . . . . And then when you try to get several programs to work together, all H@ll breaks loose.

When something like this happens, both sets of programers have to get together to think of ways around the issue. (You change your program, No you change yours. No I don't need to. You change. No mine works fine . . . .)