Computing desk
< September 6	<< Aug \| September \| Oct >>	September 8 >

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

September 7[edit]

running out of letters[edit]

Lately Firefox (before and after the latest version) does a weird thing: on some websites, ‘y’ and ’z’ do not show up at all and ‘x’ appears as ‘"’. Is it just me?

Sites where I've seen this, so far:

www.matrix.nwmls.com
www.doesnotplaywellwithothers.com
boardgamegeek.com
dieselsweeties.com
www.channel101.com
www.esperanto.mv.ru (affecting ‘z’; ‘xy’ do not occur in Esperanto)

—Tamfang (talk) 04:37, 7 September 2019 (UTC)[reply]

Sounds like some part of the memory where the fonts are stored was accidentally overwritten by something else. This could be caused by running a program that has a bug in it. If so, a reboot would likely fix it until that program is run again. If that doesn't fix it, maybe Firefox itself has been corrupted, in which case you would need to uninstall it and download again. SinisterLefty (talk) 10:32, 7 September 2019 (UTC)[reply]

Well, more than one font is affected. —Tamfang (talk) 20:01, 7 September 2019 (UTC)[reply]

One program can not access the memory allocated to another program. Each program has its own isolated memory space. Ruslik_Zero 20:05, 7 September 2019 (UTC)[reply]

@Ruslik0: Cheat Engine. Poveglia (talk) 00:13, 8 September 2019 (UTC)[reply]

@Ruslik0: Not technically correct, see shm_overview(7). If you're still on Windows there is bound to be an equivalent. Martin of Sheffield (talk) 09:20, 8 September 2019 (UTC)[reply]

There are several different methods to access the memory of another process in Windows. Cheat Engine uses the KeStackAttachProcess function to access the memory, but it is also possible to do direct memory mapping using the CR3 of the process, or hijack a thread of the target process to open the destination process and copy it to there, etc. Poveglia (talk) 10:43, 8 September 2019 (UTC)[reply]

These are malicious programs. Normal software can not cause such a defect. Ruslik_Zero 13:30, 8 September 2019 (UTC)[reply]

That is simply not true; a debugger is normal non-malicious software and it can access and manipulate the memory of another program. Its also not a defect but the way its intended to work. Something like KeStackAttachProcess is a documented feature, not a defect/bug. And calling a program that allows someone to cheat in a singleplayer videogame "malicious" is rather weird; CE only does stuff the user wants it to do while malicious software does stuff the user does not want it to do. Poveglia (talk) 13:33, 8 September 2019 (UTC)[reply]

You are implying that this user wanted to drop some letters from Firefox? Ruslik_Zero 20:34, 8 September 2019 (UTC)[reply]

@Ruslik0: No, I am not. Please re-read the above. Poveglia (talk) 00:33, 9 September 2019 (UTC)[reply]

@Ruslik0: I think I get it. Your use of the word defect was probably referring to OP's post, not to your comment from 20:05. I've tried to explain that that comment is incorrect, but I haven't attempted to answer OP's question. Poveglia (talk) 08:41, 9 September 2019 (UTC)[reply]

I am surprised that we are so sure this is a memory-corruption bug. Tamfang has given us very little information to go on - we don't know the OS, hardware, any configuration. Off the top of my head, this could be a network bug - is he using a proxy, a cache server, or anything that may be interposed? Are there plugins or extensions able to manipulate webpage data? Does the bug occur when he prints, views source, copies to clipboard, or saves the page? There are innumerable of OS- and DE- and hardware-dependent variables at play as well. Screenshots and more details would be helpful. Elizium23 (talk) 05:32, 11 September 2019 (UTC)[reply]

Indeed. The best we can say is that an unusual bug has occurred; it would be good to report it by following the official Firefox bug report guidelines. There are innumerable possible causes and it's very premature to diagnose this based on what Tamfang has provided. A cursory web-search shows that this issue is not "very common" and there is no "simple" fix for an end-user. We should direct him to file a report to the Firefox experts. We can all speculate wildly based on our previous experiences; but the truth is, we don't even know what broke, or even if Firefox application-software is the only item affected. Software debugging is too complicated to meaningfully speculate about a specific root-cause. I am frankly baffled why the early responses to Tamfang's question provided such apparent certainty about one specific and largely ill-formed hypothesis.

Nimur (talk)

You both seem to have missed the last line in my initial post: "If that doesn't fix it, maybe Firefox itself has been corrupted, in which case you would need to uninstall it and download again." I didn't speculate on how that corruption might have occurred. SinisterLefty (talk) 20:16, 11 September 2019 (UTC)[reply]

Thanks for all your remarks. I'm on a Macintosh. It's not a network issue: I've seen the effect on a local file (though, strangely, not other local files that use the same CSS). It's probably not a corruption of my fonts: multiple fonts are affected in exactly the same way, and it doesn't happen to the same pages in Safari. I removed Firefox and downloaded it anew, with no visible change. —Tamfang (talk) 22:28, 11 September 2019 (UTC)[reply]

Let's all chip in and buy @Tamfang: a new computer, that is probably easier than diagnosing this mysterious problem. Poveglia (talk) 06:33, 12 September 2019 (UTC)[reply]

Great, I'm about ready to try Linux again. —Tamfang (talk) 07:20, 12 September 2019 (UTC)[reply]

Maybe it's something malicious, have you scanned for malware ? SinisterLefty (talk) 02:57, 14 September 2019 (UTC)[reply]

Another helpful detail: only old fonts are affected: Courier, Times, Helvetica, Palatino. —Tamfang (talk) 22:21, 23 September 2020 (UTC)[reply]