What caused font "leading" (vertical space) to shrink in Firefox??

This is very strange. I have two machines, both essentially identical. One is called MDR and the other is called DFW. They both have ATI video cards (HD4850 in DFW with Catalyst driver 12.3, and HD5770 in DFW with Catalyst driver 12.8). Same exact set of fonts installed on both machines.

Yesterday, I was forced to restore a system image on one of them (MDR) from two weeks ago. No problems existed at the time I took that system image, and there's no problem now that I've restored it.

However I now have an issue with Firefox on that machine (which has just been restored) on some forum sites (including this one, SevenForums) where the "line density" of text is increased from what it used to be. In other words, although the font is the same type and size as it used to be, for some reason the blank space between each line (i.e. the "leading") seems to be reduced from what it was previously so that the lines are "denser" and closer together vertically.

The actual width of the forum window is identical on both machines, but on the problem machine MDR it looks like more horizontal space of each line is being used than the same line uses on DFW, and the lines are closer together vertically on MDR.

What could have caused this now, when it certainly wasn't the case on the system when that image was originally taken? I immediately noticed something strange after the restore, and this is the objective evidence.

Is there some way to change the "leading" on the font used for SevenForums (and several other forums as well?)?

(1) Here's a side-by-side of a few paragraphs. Note the vertical spacing difference, as well as the free-form variable pitch horizontal calculations and spacing which caused the number of words per line to differ. Note that the third paragraph takes four lines on one machine and three lines on the other.



(2) Here's an over-under of the two windows. Note that they are exactly the same width, as far as the forum frame is concerned. But the free-form variable pitch font used simply appears differently on the two machines.




Anybody have the answer?? I'd sure like to get MDR back to the way it used to look, which was identical to the way DFW looks and how MDR used to look before yesterday's system image restore.

Doesn't seem like repeating the system image restore should produce any different results, but I suppose I could just try it again (taking a current image that I can restore back to again if I have to, of the currently restored system which is strangely malfunctioning in this way).

Does anybody know the exact font which is used in SevenForums? Is this Segoe Regular 10 or something else?

Well... I have solved my own problem.

I don't know how/why the restore of the Macrium Reflect system image from 9/14 on MDR would have had any effect whatsoever on PREFS.JS for Firefox on that machine, but apparently it must have. Or, something else adjacent to that system image restore actually caused a change in PREFS.JS. Don't know.

But I'm going to do further research into this as I have time-based backups going well back and I will be able to absolutely confirm what I have now discovered as the explanation for this anomaly... which is that it was actually a change to one crucial line in PREFS.JS for Firefox which was responsible for the "leading" change on the SevenForums site, among others, because of the default font used on this site.

Anyway, just as an experiment, since both DFW and MDR are really almost identical (especially in my setup of Firefox on both machines) today I decided to just copy the PROFILES files for Firefox from my "working" machine (DFW) to my "shrunken" machine (MDR), to see if that would make the font display and "leading" on SevenForums now once again appear "more spacious". I actually first saved all of the current PROFILES contents, then deleted all of them, then copied all of the PROFILES contents from DFW to MDR, and then opened Firefox on MDR.

And, amazingly... sure enough IT DID FIX THE PROBLEM! The "leading" on SevenForums was once again back to "normal", as it had been previously before the system image restore which somehow had caused the damage.

I then did a folder/file compare of the current PROFILES contents (which was now on MDR but copied from DFW, and which was now displaying properly) to the saved original PROFILES contents from MDR (which was displaying shrunken). And of all the files which were different, it appeared that PREFS.JS would be the most likely file to reveal some critical internal difference.

So I compared PREFS.JS (normal) from DFW/MDR to the original PREFS.JS saved for MDR (shrunken). And of all the lines which were understandably and reasonably different, there was one line which seemed suspiciously relevant:

Code:

MDR shrunken: user_pref("intl.charsetmenu.browser.cache", "Big5, windows-1251, us-ascii, UTF-8, ISO-8859-1");

DFW/MDR normal: user_pref("intl.charsetmenu.browser.cache", "UTF-8, windows-1252, ISO-8859-1, EUC-KR, Big5");

In other words, it was obviously some type of font-related setting (though I hadn't actually changed any of the few basic font-related options available for user customization) found in PREFS.JS which was somehow responsible for the "reduced leading".

And of those two lines, certainly the one obvious difference which is very very suspicious, and very likely the explanation, is WINDOWS-1251 vs. WINDOWS-1252.

I believe this font must be Segoe Normal 10, which I believe is the font used by default in this forum's software. And I'll bet that the WINDOWS-1251 code page looks slightly different than WINDOWS-1252 insofar as its "leading" and horizontal spacing.

In fact, I was so convinced of my discovery that I decided to try replacing just that one line in PREFS.JS (which specified WINDOWS-1252 and was now "working normally again" on MDR) with the original line from the saved PREFS.JS (which specified WINDOWS-1251 and was causing reduced "leading"), to see if I could bring back the "failure".

And... sure enough... with just that one line put back into PREFS.JS (on MDR, but copied from DFW, and which specified WINDOWS-1252) from the saved PREFS.JS (which was on the malfunctioning MDR, and which specified WINDOWS-1251), sure enough I was able to once again bring back the reduced "leading"!!

And restoring that one line back to the one which specified WINDOWS-1252, well I was once again able to get the "leading" back to "normal".


I then restored the entire original (and failing) DFW set of files from PROFILES to confirm that the failure was back, and it was. Then I changed JUST THAT ONE LINE in PREFS.JS on MDR, to be the line from DFW... and IT DID NOT CHANGE. This was a bit surprising, but I decided that maybe there were other lines as well in PREFS.JS which were related to this but that were not "as critical", that I had overlooked.

So I then simply copied the entire PREFS.JS from DFW into MDR (with the rest of the files in PROFILES still back to where they started, i.e. they were the original MDR files not the copied DFW files) and sure enough it was now working properly again.


Bottom line: it's absolutely something in PREFS.JS which is at work here, in determining how certain fonts get displayed as far as their "leading". In particular, WINDOWS-1251 appears "reduced/denser" and WINDOWS-1252 appears "more spacious".

At least that's my theory.

Whatever the actual theory (and hopefully someone else can chime in if they know the real story here), I have now "repaired" my MDR Firefox. It has nothing to do with Windows 7, but rather PREFS.JS for Firefox, where the answer lies.

One more piece of experimental data to add to the clues and observations.

To support my theory that it was the reference to Windows-1251 (CP1251) which is the "culprit", I ran a test changing just that one single item in PREFS.JS which currently shows Windows-1252 (CP1252) and which has the proper, normal "leading" on SevenForums.

And, sure enough, with just that one single change, the display of pages on SevenForums once again "shrunk" to the more denser presentation with reduced "leading".

AHA! The smoking gun.

So, if changing Windows-1252 to Windows-1251 is enough to "break" the "leading", then putting it back from Windows-1251 to Windows-1252 should restore the normal behavior, right?

Wrong. It did NOT restore normal "leading"... even though this one line now did once again show Windows-1252. Obviously, something else must have happened when I opened Firefox with the Windows-1251 value installed, and I don't know what. So going back on this one line from Windows-1251 to Windows-1252 apparently also needed the reversion of that other line, which I had not done... and it did not automatically get done by the restored presence of the Windows-1252 value. A bit "mystical" and hypothetical, but this must be what's happening.

In my mind this is consistent with my earlier observation and statement that there is obviously at least one other line somewhere in PREFS.JS which is related to fonts and "leading", but which I haven't discovered, but which can force the display behavior as if it were still Windows-1251 even though that one particular line actually has Windows-1252 in it again.

So once again, I re-copied the ENTIRE version of PREFS.JS over from DFW to MDR and now sure enough once again the normal "leading" from Windows-1252 was back in effect on SevenForums.


I'm sure if I pursued this further, I'd find that other one (or possibly several) font-related lines in PREFS.JS which needs to be changed in conjunction with this primary "intl.charsetmenu.browser.cache" statement.

It is 100% guaranteed that the presence of Windows-1252 on this line produces "normal leading" and Windows-1251 produces "reduced leading".

Nevertheless, my problem is solved. I've got SevenForums (and other forums which also use Segoe Normal 10 font) back to "normal" in Firefox on MDR.

Case closed.

Discovered where the particularly relevant PREFS.JS line appears in Firefox:

View -> character encoding

It's not actually a normal "options" setting, although it can be modified using "Customize list...".

The "normal" presentation of SevenForums in Firefox for me is as follows (which apparently does NOT use Windows-1252):

Well, now I'm getting crazy.

Turns out that the "proper spacing" between lines only remains in effect for the very first Firefox session during which I visit a site (e.g. SevenForums) where the "character encoding" is Western (ISO-8859-1).

But something apparently then happens to the Firefox settings during the visit to that page, which then gets saved when I close Firefox. And the changes that get saved revert the presentation to the "reduced leading" situation.

So the next and subsequent times I open Firefox and re-visit that same site, well now I'm right back where I started... with the shrunken "leading" and denser text.

I can re-copy PREFS.JS from DFW to MDR, and that will "fix" the problem so that the next time I open Firefox it will once again be "normal leading". But if I visit any forum whose "character encoding" is Western (ISO-8859-1), sure enough something happens and I've now been clobbered.

And actually, I don't even need to close and re-open Firefox for that "clobbering" to occur. All I have to do is visit another page, and then come back to the original forum page with that Western (ISO-8859-1) encoding... and now my "leading" has been reduced.

It's the mere VISIT to that other forum page which "does the damage". As long as I stay in SevenForums and view any number of threads and forums, each of the displays remains "good". But if I go to a new forum, I've been clobbered. And if I come back to SevenForums from that other forum, I'm clobbered.

In other words, the values in that "intl.charsetmenu.browser.cache" item still shows exactly what it was when I copied it from DFW to MDR and which caused that initial visit to the forum page to display correctly. It's still got Windows-1252 (which I'd assumed was the answer to all of this behavior), not Windows-1251, but apparently that turns out not to be all that's relevant.

There's something else at work here. I'm going to have to compare PREFS.JS "after closing that first Firefox session" to what I freshly copied, to see what else is changing.

Ok. I may have conquered this one.

I restored ALL of the files in my PROFILES folder for Firefox on MDR from a 9/18 backup (for MDR, not DFW). I know things were certainly working fine on 9/18, and I hadn't restored a system image on that date but was continuing to use my normal ongoing system. Yes, VSS wasn't working for me with NovaBackup (which ultimately prompted me to restore a "system image" earlier this week, which started all of this "leading" problem), but certainly Firefox was working fine. So presumably the relationship of ALL of the files in the PROFILES folder must have been internally consistent and 100% correct on that date, so restoring ALL of them (from MDR) TO MDR should have just put me back to 9/18, as internally consistent and fully-functional as everything was on that date.

So, since there's really nothing changed in my Firefox setup over the past few weeks I have no problem just restoring the PROFILES folder from a few weeks ago, and 9/18 was as good a first try date as any.

I did NOT copy PREFS.JS from DFW, nor did I modify it from whatever it had in it on 9/18. I simply restored the entire PROFILES folder for MDR that got backed up on 9/18.

And remarkably things are now seemingly working "normally" again. The SevenForums pages are displaying with proper "leading", as are the other forum sites which also use Western (ISO-8859-1). And I can go back and forth between forums and no "clobbering" is done. And I can close and re-open Firefox, and the proper "leading" is retained.


So, I'm kind of baffled. I don't like "numerology" solutions, and just keep trying something until it "works" and then try and explain WHY it worked with no real objective understanding of truly why it worked.

But I've spent lots of time over the past few days on this one, ever since that system image restore earlier in the week obviously did something... although it was simply supposed to be a restore of a static system image from 9/14 of my fully operational and 100% normal system on that date.

Nevertheless, this last recovery action... of restoring the 9/18 PROFILES folder for Firefox from MDR to MDR... well, it seems to have done the trick (at least it seems that way for now).

Fortunately I have a daily INCREMENTAL backup regimen that runs automatically and that gives me up to 90 days of "daily recovery capability" for anything on my system, and 6 months of "monthly recovery capability" (i.e. if something existed on the 1st of that month it got backed up on that 1st-of-the-month's FULL backup), so I had the luxury of even trying this recovery action which I honestly could not have given much hope of solving my problem.

And yet... it does (for now) seem to have solved the problem.

Fingers crossed.

I have now found 100% what the cause of the "reduced leading" problem is.

It IS in PREFS.JS, but it is NOT the "font" item I thought was central.

Instead it is the two lines:

user_pref("gfx.blacklist.direct2d", 3);
user_pref("gfx.blacklist.layers.direct3d9", 3);

which are generated on my particular machine because I have the Flash Plugin (v11.4.402.265) enabled. If I disable the Flash Plugin these two lines are NOT generated into PREFS.JS when I close Firefox. If I enable the Flash Plugin again, when I close Firefox these two lines ARE generated.

I believe that my home page CNN.COM uses the Flash Plugin, so that every time I open Firefox I am invoking that plugin... if it's enabled.

If I delete these two lines and launch Firefox, it comes up with "normal leading". If those two lines are already present in Firefox when it is launched, it comes up with "reduced leading".


I have now disabled the Flash Plugin on this machine.

I've started another thread on this subject, so please if you know anything about the Flash Plugin or these two particular lines and what they mean or are caused by or why they exist, please post your information on that other thread.