Posts tagged ‘Mozilla’


What to do when Firefox for Windows Vista displays no text

28.12.2014

During the six hours wasted with Ubuntu today (13 would no longer upgrade, so I removed it and decided to start afresh with 14—big mistake, since it would not let me use the same hard drive), I had to open up my five-year-old Windows Vista laptop and upgrade my Firefox. After all, what were the odds that Mozilla would cock up its flagship browser on two OSs? After all, it’s fine on Mac OS X and Linux.
   As it turns out, pretty high. Just as in Windows 7, Firefox for Windows Vista displays no text. And unlike Windows 7, which was solved by switching on hardware acceleration, Windows Vista proved a bit of a bugger to fix.
   During the months where I was trouble-shooting, and after my last post, one of the more knowledgeable Mozilla volunteers admitted that there is a fault with the Cairo rendering engine in Firefox: ‘This means that (at least in your case) the issue is most likely specific to the cairo drawing backend. Good to know, thanks.’
   It is still definitely related to the 2011 bug I filed where PostScript Type 1 fonts were incompatible with Firefox due to something breaking that time.
   Firefox for Windows Vista’s bug, as far as I can make out, is down to Type 1 fonts being incompatible with the browser, even though they are compatible with nearly everything else on the OS. This is slightly different from the Windows 7 fault, as I still have PostScript Type 1 fonts on that computer, but Firefox simply ignores those when specified in a stylesheet in favour of what it can load under hardware acceleration (usually the default).
   Despite my updating some of the system fonts that were particular to my Vista set-up to OpenType (which Firefox might have trouble with sometimes, too), that did not fix it. Firefox requires you to delete fonts off your system.
   On some websites, including Facebook, Helvetica is specified before Arial in stylesheets. If your Helvetica (not Neue Helvetica) is PostScript Type 1—and it probably would be on a Windows machine—Firefox will detect it, and return blank spaces.
   This is still a daft state of affairs with Cairo. Here’s how (to my very basic layman’s mind, and obviously to the minds of everyone at Adobe and a bunch of other places) how a program should deal with fonts:

* Is it installed on the system? Yes.
* Use it.

Firefox seems to adopt this approach:

* Is it installed on the system? Yes.
* Let’s ignore the ones our programmers dislike in favour of the ones our programmers like, which would only be certain TrueType fonts, and to heck with the people who have licensed other fonts and installed them in good faith. Let’s punish anyone who decided to carry over older software. Let’s also fail selected OpenType fonts such as the italics in Source Sans Pro for no apparent reason. [PS.: If the first font family is incompatible, let’s display nothing. On a stylesheet, if one does not work, we won’t load the second one, but we will try to load the system font even if that is incompatible, too.]

   When it comes to stylesheets, neither OS makes much sense. Normally a program would go through each font specified, and display in the first one available. I don’t understand the rationale but Firefox will skip the ones in the stylesheet even when installed, even when compatible, and opt for system fonts or those specified as defaults in the program.
   All of this is counter-intuitive, and if it weren’t for what must be my OCD, I’d never have found out, and have given up to use another browser.
   Not that IE11 is much good:

Loading up the next ASCII character makes little sense, either.
   Firefox isn’t unique in mucking up type on Windows. Back in the days of versions 1 and 2, I avoided it because it was incapable of displaying ASCII characters above 128 in the same font. This still afflicted Opera, the last time I saw it in 2010. Chrome wasn’t much better: it will pick one character and display that in a different font or fail on the font-face spec in HTML. For years, Internet Explorer would only let you use TrueType.
   I don’t fully understand why Windows browsers must behave differently—no doubt it’s to do with Flash or HTML 5 or something connected with rendering—but it is very annoying when every other program I have gets this right.

Tags: , , , , , , , , , , , , ,
Posted in internet, technology, typography | No Comments »


Firefox, Waterfox, Cyberfox displaying no text? Do the opposite of what you are told

24.12.2014

After months of avoiding the latest Mozilla Firefox because it would display no text, installing, removing, and reinstalling an older version of Waterfox just so I could do some work, and experimenting over the last day with Cyberfox, which included editing fonts, looking at GFX settings, editing the registry, and doing an awful lot of research, I have now fixed the problem of having no text in these browsers.
   As I discovered years ago, the trick is to do the exact opposite of what the experts suggest.
   When trying to set up the office network in the mid-2000s, the only way I could get it going was to do the exact opposite of expert advice, by making sure the speed on every device did not match.
   Tonight, the solution was as simple as pie. Almost every piece of advice I had received when reporting this issue was: turn off hardware acceleration. It was already turned off, so, logically, I kept looking at other things. It got to the point where I was advised by Loic, one of the helpful guys on the Mozilla forums who had hitherto walked me through possible solutions, ‘As you are able to reproduce it, could you use the tool mozregression to find a possible regression range, it will help to narrow the underlying regression.’
   Software people think I am a lot smarter than I really am, and I had to admit to the writer that I am a layman and I had no idea what he was talking about.
   The correct solution, if you want Firefox, Waterfox and Cyberfox to display text where there was none, is to turn on hardware acceleration.
   You can imagine my feeling right now: a sense of satisfaction knowing that I am running the latest, most secure browser and that I overcame this rather serious bug, with the usual disappointment in realizing that I trusted again in expert advice that wasted time for a lot of people, including those kind developers on the Mozilla forums, and Alex, the guy behind Waterfox, who were trying to find a solution for me. I simply do not know enough about computers and software beyond what I have to do to make a living.
   The concluding remark from one of the guys, Nicolas, on the Mozilla thread was, ‘This means that (at least in your case) the issue is most likely specific to the cairo drawing backend. Good to know, thanks.’
   I hope documenting this bug has helped someone out there. Merry Christmas!

Tags: , , , , , ,
Posted in internet, technology, typography | 4 Comments »


Cyberfox day two, or, the day it, too, stopped displaying text

24.12.2014

Rather than repeat the story in new words, here is a draft of the post that was sent to Cyberfox’s support forum.

The short story: Cyberfox no longer displays text as of this morning after working well for its first evening yesterday after installation for the first time. Glyphs that are not from a @font-face linked font will not show, so if a page is calling fonts from the system, the browser displays blank text. Nothing happened overnight. I switched the machine off, and when I switched it on again, Cyberfox exhibits this behaviour.
   The long story: in 2011, Firefox had a bug which meant there was no backward compatibility with PostScript Type 1 fonts (https://bugzilla.mozilla.org/show_bug.cgi?id=628091). This is very similar to that except the text areas are blank, rather than filled with squares or hex codes.
   About two Firefox versions ago (I am guessing v. 32), the browser stopped showing text. I switched to Waterfox, which lasted one more version before it, too, stopped showing text. I downloaded Cyberfox last night and was truly pleased that here was a Firefox-based browser that actually worked. Text displayed as normal, and these were my Type 1, TrueType and OpenType fonts. To top it off, Cyberfox’s rasterizer and the way it handled sub-pixel rendering was superior to that of the other two browsers (see my blog post at http://jackyan.com/blog/2014/12/switching-to-cyberfox-after-waterfox-and-firefox-stopped-displaying-text/ for two screen shots of the type). Naturally, I was over the moon and made Cyberfox my default.
   Just to be on the safe side, I turned off hardware acceleration as when I posted the above bug to Mozilla Support, I was told that that could be a culprit. I made no change to OMTC.
   Today, as mentioned, Cyberfox has become just another Firefox where no text is displayed. But the really weird thing is that the typography, for the type that does show, is identical to Firefox and Cyberfox: the superior rendering is gone.
   Also, I’ve since altered the font family I use as a default for Windows displays to OpenType (I work in fonts), so there should no longer be an backward-incompatibility issue. Nvidia updated one of its drivers today, so I let that happen, and confirmed that all my drivers are up to date.
   Reinstallation (while keeping profile data) actually fixes everything: the type is back, rasterized more sharply,
   I was using Australis as the theme but have since gone back to classic.
   I’d be grateful for any light you can shed on this as I’m keen to stay within the Firefox 64-bit family. Whatever makes Cyberfox display better than the other two immediately after installation (though not after a reboot) solves this major problem of no type appearing.

   The different rendering method is the fix. The questions are: why does Cyberfox render type differently if it’s Mozilla Firefox-based? And why does rebooting my computer change this setting?

Tags: , , , , , , , , , ,
Posted in internet, technology, typography | No Comments »


Switching to Cyberfox, after Waterfox and Firefox stopped displaying text

23.12.2014

Since the Firefox for Windows updates in November, I’ve had a big problem with the Mozilla browser, and the Waterfox 64-bit version based on it: they won’t display text. I had to downgrade to Waterfox 32.0.3 for the last month or so, but it’s begun crashing more and more regularly (from once a day to thrice today—I visit largely the same sites, so why does software “decay” like this?).

   On the latest incarnations of Firefox and Waterfox, linked fonts work, but the majority of system fonts vanished from the browser. And, for once, I’m not alone, if Bugzilla is any indication. It is probably related to a bug I filed in 2011.
   I’ve had some very helpful people attend to the bug report—it’s great when you get into Bugzilla where the programming experts reside—but sadly, a lot of the fixes require words. And, unfortunately, those are the things that no longer displayed in Firefox, not even in safe mode.
   As many of you know, there’s no way I’d switch to Chrome (a.k.a. the ‘Aw, snap!’ browser) due to its frequent crashes on my set-up, and its memory hogging. There’s also that Google thing.
   After some searching tonight, I came across Cyberfox. It’s not a Firefox alternative that comes up very often. Pale Moon is the one that a lot of people recommend, but I have become accustomed to Firefox’s Chrome-like minimalism, and wanted something that had a Firefox open-source back end to accompany it. Cyberfox, which lets you choose your UI, has the familiar Firefox Australis built in.
   I made the switch. And all is well. Cyberfox forces you to make a new profile, something that Waterfox does not, but there isn’t much of an issue importing bookmarks (you have to surf to the directory where they are stored, and import the JSON file), and, of course, you have to get all your plug-ins and do all your opt-outs again. It also took me a while to program in my cookie blocks. But the important thing is: it displays text.
   You’d think that was a pretty fundamental feature for a web browser.
   The text rendering is different, and probably better. I’ve always preferred the way text is rendered on a Macintosh, so for Cyberfox to get a bit nearer that for some fonts is very positive. It took me by surprise, and my initial instinct was that the display was worse; on review, Firefox displayed EB Garamond, for example, in a slightly bitmapped fashion; Cyberfox’s antialiasing and subpixel rendering are better.

Firefox and Waterfox on Windows 7
Firefox_Screenshot_2014-12-23T14-14-06.693Z

Cyberfox on Windows 7
Firefox_Screenshot_2014-12-23T14-13-23.809Z

Here’s where the above text is from.
   Gone is the support for the old PostScript Type 1 fonts (yes, I still have some installed) but that’s not a big deal when almost everything is TrueType and OpenType these days.
   The fact Cyberfox works means one of two things: (a) Cyberfox handles typography differently; or (b) as Cyberfox forces us to have a new profile, then there is something in the old profiles that caused Firefox to display no text. That’s beyond my knowledge as a user, but, for now, my problems seem to be solved—at least until someone breaks another feature in the future!

PS.: That lasted all of a few hours. On rebooting, Cyberfox does exactly the same thing. All my text has vanished, and the rendering of the type has changed to what Firefox and Waterfox do. No changes to the settings were made while the computer was turned off, since, well, that would be impossible. Whomever said computers were logical devices?
   Of yesterday’s options, (a) is actually correct—but how do we get these browsers behaving the way they did in that situation? In addition, the PostScript Type 1 fonts that the browser was trying to access have since been replaced.

Tags: , , , , , , , , , , , ,
Posted in internet, technology, typography | 3 Comments »


Water trumps fire

09.02.2014

Since I used to post updates of the web browsers I used: I have switched to Waterfox, replacing Firefox.
   Since the latest Flash updates a few weeks back, Firefox has been crashing twice a day. Other weird things have happened, too, like the save file dialogue box failing to appear after several hours’ use, or the mouse pointer flickering like crazy.
   I also haven’t had Waterfox change pages on me automatically, a bug that has been with Firefox for years but remains unsolved.
   Firefox for Windows is not designed for 64-bit machines, but Waterfox is. Since changing browsers, I have had a crash-free existence.
   It’s not the first time I downloaded Waterfox but abandoned it last time. I can’t remember the exact reasons but it would have been either losing some of my settings, finding that its speed was worse than the 32-bit version, or its high memory usage.
   The last of those three still holds true—Waterfox will eat through over a gig of RAM—but everything from Firefox comes across perfectly and it is slightly faster.
   Sadly, I have had to remain on Firefox for my 32-bit laptop running Windows Vista, where it has been crashing regularly since the last Flash update.
   I’m still on Firefox on Ubuntu and Mac OS X, but it looks like there is some major issue with Firefox and Flash when it comes to Windows. This is not the first time, either, but it is enough to have me stay on Waterfox for the foreseeable future.

Tags: , , , , , , , ,
Posted in internet, technology | No Comments »


Chromium remains buggy; and I get charged twice for parking (thrice if you count my rates)

18.03.2011

I am happy to say that Firefox 4 Release Candidate 1 is working smoothly with no crashes to date. It reminds me of, well, Firefox 3·0, before Mozilla started doing weird things to it and we had the multiple-crashing 3·5 and 3·6. Let’s hope this situation lasts.
   Meanwhile, the bugs I reported to the Chromium people in October and November 2010 have finally received responses. It’s too long compared with Mozilla. I told the chap that I had given up on Chrome, but I downloaded it today just to see where things were at.
   Based on the latest Chromium, the incomplete font menu bug has, indeed, been fixed, though various font-changing ones still appear present. There are still font-linking and character-set issues. (The images below have had their colour depth reduced for faster loading.)

Chromium 12
Chromium 12Above and left: When Chromium hits a ligature, the line or part of the line changes font. Opera does something similar: it changes the font of the one word that contains the ligature.

Chromium 12Left: Chromium might just decide to change fonts anyway—likely a Postscript error already sorted in Mozilla thanks to the likes of Jonathan Kew.

Chromium 12
Above: The font-linking problem on the home page of Lucire still has not been solved. There is no problem on IE8, Firefox or Opera. I can’t report on IE9 as my psychic powers are not strong enough to determine what is being told to me through the heavenly dimensions.

Chromium 12
Above: This one paragraph is properly linked—what causes it to work and the others not to is unknown.

Chromium 12
Above: Go outside the regular Latin set, and Chromium falls all to pieces, just as it always did.

Chromium 12
Above: At least there are fewer font changes than last time—though Chromium continues to struggle with soft hyphens.

   Meanwhile, after I reported spam faxes (a breach of the Telecommunications Act) to Telstra Clear, I was surprised to learn that my case was never examined. I had to open a new ticket with new faxes today. The excuse was the backlog of work post-Christchurch earthquake and, in the circumstances, I had to accept that.
   One was for a law firm, as far as I could make out. I wouldn’t hire a lawyer who breaches the Telecommunications Act. They shot themselves in the foot with that one.
   My main reason for calling, however, was the Text-a-Park service that the WCC offers. I hate cellphones, but had brought one with me on one of those rare occasions, and decided to give the service a go. I fed in the code, dialled 7275, and was told by the meter that the transaction had failed. No parking ticket was generated.
   Just as well. I prefer to use a credit card anyway, and fed that in. I got my ticket and my credit card was charged.
   Problem: as I walked away from my car, I received an SMS saying (sic), ‘Thanks for using TXT-a-Park. Your transaction for $6.50 has been accepted.’ I have no idea how one gadget says it’s failed and another says it’s succeeded, and my quantum physics isn’t good enough to figure out into which alternative universe this supposedly successfully printed ticket went to.
   Of course, the charge appeared on my Telstra Clear bill today.
   There’s a reason that jokers like me don’t use cellphones. Because, each time we do, they bite us on the bum. Though a buttcheek bite is better than testicular cancer.
   I’d urge folks to check their bills—if you haven’t received a ticket from a Wellington parking meter, and you still got charged for it, then give the telco a call.

Tags: , , , , , , , , , , ,
Posted in business, internet, New Zealand, USA, Wellington | No Comments »


Microsoft Internet Explorer 9: the worst browser on the scene

15.03.2011

Microsoft has released its Internet Explorer 9 to much fanfare at SXSW. I’m really not sure what the fuss is, because it appears, as usual, the browser hasn’t been tested.
   Here it is on my Asus laptop, running Vista.

IE9

That’s apparently my company’s home page. Looks slightly different to how Firefox, Chrome and Opera display it:

Firefox

   I might dislike Chrome but at least that browser shows something other than pitch black with a few tiny details.
   Let’s go to the most well known website in the world. Surely IE9 can display that and that its beta testers must have been to Google. Unless Google is banned at Microsoft and everyone uses Bing. Here’s what Google’s home page looks like:

IE9

I knew Microsoft was aiming for a minimalist look, but isn’t that taking it a bit far?
   You won’t see it on the screen shot above but there is a blinking cursor. You can begin typing, but nothing echoes on the screen. On pressing ‘Enter’, you do get a search page, and, lo and behold, it resembles the usual Google results’ page—kind of.

IE9

   What if I scroll down?

IE9

   Conclusion, based on one machine that can run every other browser: Microsoft Internet Explorer 9 is a load of cobblers. I managed to crash it twice on the first two web pages I visited, within the first two minutes. The rest, you see above. I couldn’t be arsed doing more with it.
   Mr Gates, if you want to come back to me when your team has actually tested your browser, I will be happy to give it another shot.

Tags: , , , , , , , , , ,
Posted in business, internet, technology, USA | 15 Comments »


Hopefully the last Firefox 3 blog post

05.03.2011

Since discovering that Firefox 4 Beta 13 is stable, I have spent less time with Firefox 3·6, the buggiest, most oft-crashing program I have ever used in 30 years of computing.
   But I used it today enough times to net myself five crashes, though this is above average. The ‘unmark purple’ bug that plagued me for so long has disappeared, which indicates it was an error with an extension (Flash, maybe?), and the average of four per day has decreased to two to three (on the days I use Firefox 3·6 exclusively).
   However, since the ’quake, I have still netted a number of errors, and apart from one, there is no pattern to them. Here are the last 13 on this machine (I’ve used it a bit more on my laptop, which doesn’t have 4 Beta):

1 × [@ nsTArray::IndexOf >(nsAppShellWindowEnumerator* const&, unsigned int, nsDefaultComparator::RemoveObject(imgCacheEntry*) ]
1 × [@ InterlockedCompareExchange ]
1 × [@ PR_AtomicDecrement | nsSupportsCStringImpl::Release() ]
1 × [@ hang | mozilla::plugins::PPluginScriptableObjectParent::CallHasProperty(mozilla::plugins::PPluginIdentifierParent*, bool*) ]
1 × [@ hang | ntdll.dll@0xe514 ]
1 × [@ nsRuleNode::WalkRuleTree(nsStyleStructID, nsStyleContext*, nsRuleData*, nsCSSStruct*) ]
1 × [@ WrappedNativeProtoMarker ]
1 × [@ F_592283983_____________________________________________ ]
1 × [@ nsExpirationTracker::RemoveObject(gfxTextRun*) ]

   I have no idea what any of this means, but to the layman it suggests the gremlins are everywhere in the program. (The defence by Firefox proponents in claiming that post-3·5 versions are the most stable releases falls on deaf ears here: 3·0 and 3·6·10 crashed far less often.)
   I’ll sure be glad when Firefox 4 rolls out, and I have been really impressed by the bug-fighting and beta-testing programmers. They have actually listened to what I have to say and confirmed that most of the bugs I have reported existed. It’s already a darned sight better than Chrome and its nearly-every-session ‘Aw, snap’ pages, of which no screen shot can be taken.
   But based on the above crashes, there is, of course, no mystery on why Chrome’s market share has increased and Firefox’s has decreased. Chrome crashes, but not as often—and most won’t care about its typographic problems or the lack of support. Mozilla needs to get 4 out ASAP: the more 3 crashes—and judging by the comments in Bugzilla, the rate of crashing remains remarkably high—the more likely users will hop over to the competition.

Tags: , , , , , , , , , , , ,
Posted in business, design, internet, technology, USA | No Comments »


Firefox 4 Beta 13 passes my tests

24.02.2011

Firefox 4 Beta 13 works, and I have not found any bugs with it.
   I may be wrong, but I believe this is the last beta before release.
   What’s amazing is that the bugs I have been complaining about for a long time have each been fixed. In other words, the reporting system works.
   While for many versions, most of the Beta 4 text was unreadable, eventually bug reports to both Mozilla Support and Bugzilla got things on the radar.
   That took a bit too long for my liking, and you do have to persist. But once I was “in the system”, things got resolved fairly quickly.
   One of the Mozilla boffins created a patch that I could use to tell him what fonts I was using, to trouble-shoot the unreadable UI.
   When those font issues were fixed, I noticed that there were still some errant numerals—a bug that Chrome also has. The difference: at Mozilla, it got fixed. Someone (Jonathan Kew) believed me, had at the back of his mind what it was, and wrote code to sort it out.
   We all worked it out together, with a layman like me providing screen shots and some public domain fonts on request, and the real experts then doing the hard yards.
   The main thing was that I was believed and it was confirmed, on each occasion, that I had a valid complaint.
   Unlike a certain other browser from a company which, I must say, did a good job with the Google Person Finder in the wake of the Christchurch earthquake.
   I don’t deny they do good sometimes—it’s just that they slip up far too often other times.
   The Chrome bug reporting and forums are about as useless as those for Blogger.
   Features I’m discovering in Beta 13 are really nice, now that I am no longer being distracted by the wrong fonts displaying.
   The box in which I am entering this text can be resized—not something I could do on Chrome or Firefox 3.
   More fonts’ kerning pairs are being read (see above left): someone at Mozilla likes typography. Some text-sized pairs look a little tight, but that’s a small complaint.
   Some alternative characters in OpenType fonts are showing up—whether that was intended or not, I don’t know. But it seems Firefox 4 is, at least, accessing them.
   It’s not a memory hog: I estimate the memory usage is on a par with Firefox 3.
   The promise of Firefox being reliable seems to have been realized: it took me days to crash Beta 12, and Beta 13 is so far, so good.
   The user interface is cleaner—not Chrome-clean, but pretty good.
   The speed seems improved, though I still feel Chrome is quicker. But I’d rather wait the extra hundredth of a second and have the page displayed properly.
   Hopefully, once installed on my system, Firefox 4 is going to work a treat. Well done, guys.
   If you’re going to have speedy R&D, it sure pays to have a system which embraces user experiences, working as much in parallel with your own team as possible.

Tags: , , , , , , , , , , , , , , ,
Posted in business, culture, design, internet, technology, USA | 2 Comments »


Type-changing bug identified—not that it matters next to Christchurch

22.02.2011

It’s quite pathetic to be blogging about something like this on the day of the Christchurch earthquake, but Jonathan Kew, who has kept on the font-changing bug in the Firefox 4 betas after I mentioned it to him, has created a patch that sorts the problem out. Apparently, it applies to old PS1 fonts: Firefox was rejecting the glyph index 31 in these fonts.
   Jonathan is a real ally to the type community, and understands the industry’s needs very well. We’re lucky to have a guy like that involved in browser development. Here’s hoping for approval for the patch.

I’ll repeat parts of what we wrote on the Lucire site today: ‘New Zealand Red Cross is accepting donations …
   â€˜Twitter updates can be found at hashtag #eqnz.
   â€˜Google has a Person Finder for those who are looking for people or wish to report they are OK.’

Tags: , , , , , , , , ,
Posted in design, internet, New Zealand, technology | No Comments »