Help! Downloaded update now Scrivener won't load

[JimRac]

This is ridiculous. This thread alone is almost a YEAR OLD. We are paying customers. And either we have to spend hours fiddling around with things and potentially making it worse, or we have to use an older, unpatched version which may not be secure. Not even to respond is just unacceptable.

I've PMed all of the moderators of this forum that accept them.

Hi,

While L&L tech support does show up, these forums are intended more for user-to-user peer support.

If you want to connect directly with L&L Windows technical support, email them. URL below. They have always responded to me within a day or so.

https://www.literatureandlatte.com/support.php#section-email

Best of luck, hope you get your problems resolved.
Jim

[glaedr]

This is the technical support forum on the official forum, why is being left to people without guaranteed training?

[JimRac]

If you review the posts in the forums, I think you'd see that many questions are answered and issues resolved user-to-user. That's a reasonable support model for a tiny company like L&L.

The support page lists various ways to get help, but in my opinion it's fairly clear what to do when all else fails: "If you have exhausted the other channels and still can't find the answers for which you are looking, you can e-mail support directly. "

https://www.literatureandlatte.com/support.php

Again, I do hope you get your issue resolved.
Jim

[glaedr]

I can understand relatively minor issues being fixed by peers, but this is quite a big thing affecting multiple people. Somebody had already contacted them without much success. Personally, I've been almost entirely without time over the past few months, hence popping back in and asking if there had been any updates. And in fact, when I got back, I did send messages. The fact remains that if they've been made aware of the issue, either through that email or this thread, it really should have been fixed in subsequent updates after this amount of time.

[JimRac]

Sorry glaedr, I missed that you had already emailed L&L support directly. That was the only point I was trying to make.

[devinganger]

I can understand relatively minor issues being fixed by peers, but this is quite a big thing affecting multiple people. Somebody had already contacted them without much success. Personally, I've been almost entirely without time over the past few months, hence popping back in and asking if there had been any updates. And in fact, when I got back, I did send messages. The fact remains that if they've been made aware of the issue, either through that email or this thread, it really should have been fixed in subsequent updates after this amount of time.

Maybe I'm missing something, but the ultimate cause(s) of the problem aren't due to the software, they're due to specific types of misconfigurations/corruptions in the user's profile/configuration interacting with code provided by the Qt4 framework. As a result, there is no one general fix that can be released without introducing other problems -- each affected user needs to be worked with directly to find and resolve the issue.

I realize that's not what you want to hear, but it's how it is. This isn't a case of an issue being left for peer support -- it's a case where the fix requires one-on-once cases with support. And in the early months of the issue, nobody (including support) had any issue of where the conflict was. Some people contacted support and didn't want to do the troubleshooting or use any of the workarounds provided (using a new user profile, etc.) -- what can you do at that point?

[kewms]

Maybe I'm missing something, but the ultimate cause(s) of the problem aren't due to the software, they're due to specific types of misconfigurations/corruptions in the user's profile/configuration interacting with code provided by the Qt4 framework. As a result, there is no one general fix that can be released without introducing other problems -- each affected user needs to be worked with directly to find and resolve the issue.

And if the underlying issue lies outside of Scrivener, there's a good chance the fix does, too.

Katherine

[glosfor]

And if the underlying issue lies outside of Scrivener, there's a good chance the fix does, too.

If you are referring to the "Warning: QSettings: RegEnumValue failed" problem, you can see (what I believe to be a version of) the source that is causing this here (scroll down to line ~248).

The fix is relatively straight forward. As I mentioned back in February:

The registry is basically a database, and any application running on the computer might be adding or removing keys at any time. Getting a 'count' and then walking 'count' keys runs the risk of someone having make a change between the two operations. A better approach is simply to keep calling RegEnumKeyEx, incrementing the requested index until ERROR_NO_MORE_ITEMS is returned.

While it might be interesting to know specifically what is causing this, perhaps it would be sufficient to repair it. At a minimum, it seems like a check for that specific return code in this routine should suffice:

Code: Select all

if (res == ERROR_NO_MORE_ITEMS)
   break;

[glaedr]

But it isn't happening with any other software? It's Scrivener that we're all having the issue with. And it's from a particular update. So though it might be something outside of the software that it's clashing with, wouldn't it be better to find a different way to code whatever it is that's clashing rather than have to go through the same lengthy process with who knows how many people?

I've never seen an error message, it just won't load. So I'm not sure if it's that one or not. Thanks for the link though, I'll take a look.

...Nope, still not working.

[LAP]

This problem appears to be related to how Scrivener parses the contents of certain font files. Clearly, for some users, there are issues with specific fonts. Our best guess currently, is that Scrivener can't parse certain fonts, or its crashing when parsing malformed fonts and our parser needs to be adjusted to fail gracefully when parsing malformed fonts - much like our RTF parser has to when parsing imported and malformed RTF.

The only way we can fix this is if we can duplicate the issue. We tried unsuccessfully in the past to duplicate, but were unable to. Given all the focus on Windows Scrivener 3 development and preparing for beta - this issue got lost in the piles of things to do - we are sincerly sorry for that. However, to fix this please help us by providing a step by step use case where we can duplicate - specifically we need to look at a log file and see what is the last font Scrivener tried to load and choked on, without this kind of information we can't fix this issue. No doubt the fix will be simple and quick once we know what needs to be fixed.

The Gungsuh font, seems to be a Microsoft font, which is installed on some Windows distributions. We found it and downloaded the 129 USD font and it ran without issue; however, we noticed other versions from other sites (i.e. the free kind), and discovered that the font was often corrupted; however Scrivener still did not crash on these malformed download fonts, but there may be lots of other download sources we have not tried that do crash scrivener; so, once again we're at a loss here.

The issue regarding failing to start Scrivener for no obvious reason, is also without a reproducible use case, this is like chasing ghosts. Process Monitor files do not show sufficient information to help us trace this further.

Lee

[glaedr]

All I did personally was to accept the suggested update. It wouldn't reopen after that. Ran as admin, just in case and nothing. Un/Reinstalled, including removing traces and same problem with a clean install. Removed all again and reinstalled the older version and it opened again. So it's not the save files, nor did removing all but the regular Windows fonts work by the way and I've just checked and don't appear to have Gungsuh at all. I ran virus/malware scans, nothing found. No issues that I've noticed with anything else. I've also tried three or four times to install the update again, - most recently last week - in case the download had been corrupted with no change.