all scripts have disappeared

[chrizilla]

@erosman : Thank you for responding so quickly to issue #39 . I will reply there as soon as I can. But I am facing even bigger problems now:

• Following your advice, I added the linter rules in Firemonkey options, but saw now effect.
• Therefore I restarted Firefox.
• After restart, ALL my userscripts have disappeared from Firemonkey !!
  The page is gray and there is a zero counter in the left bottom corner.
  My history log is still there though, but all scripts are gone.

It looks like adding the rules entry has corrupted the storage database. Because I have been using Firemonkey intensively for months and never experienced any bugs. The data loss occurred immediately after adding the custom lint rules.

I am desperate. Can you please help me ?
I keep the Firefox window open and won't touch anything for fear of making the problem worse ...
I hope you can quickly get back to me with any help. Thank you so much.

Firemonkey 2.74 on Windows.

[chrizilla]

I can now definitely confirm that the local storage got corrupted (very likely due to adding the linter rules, as that was when the corruption occurred):

• The export button in the Firemonkey options does not work.
• The export button in the Firemonkey sidebar script list (on the left side) exports 14 scripts despite
  • the list appearing empty
  • counter says 0 (zero)
  • the actual number of my scripts being 50

[erosman]

Have you been using FM 2.74 all the time?
FM2 uses CodeMirror and the linter rules are different, while FM3 uses Monaco-editor and ESLint.

• Go to FM options page
• Open Dev Tools (F12) -> Console
• Refresh the page
• See if there are any errors

While on the Dev Tools, paste the following in the console tab and then ENTER. It would show storage content

browser.storage.local.get().then(console.log).catch(console.error);

[chrizilla]

Many thanks for looking into this.

@erosman :
Have you been using FM 2.74 all the time?
FM2 uses CodeMirror and the linter rules are different, while FM3 uses Monaco-editor and ESLint.

I never used FM3.
I am not 100% certain, but relatively sure I have not updated FM in over a year.
about:addons says:

Author: erosman
Version: 2.74
Last Updated: May 2, 2025

I don't know if that's when I or you last updated the addon, but either way it was not updated on my device since at least 2 May 2025, apparently. This is also in line with how I remember it.

[erosman]

FM3.0 is not released yet. It is in beta.
The reason I asked was that in #39 you mentioned ESLint which is in FM3 and not in FM2.
The example I gave was for ESLint and FM3.

FM2 uses CodeMirror and JSHint for linter.
To do the same in FM2.74, look at the example in the included Help under CodeMirror.

[chrizilla]

@erosman :
* Go to FM options page
* Open Dev Tools (F12) -> Console
* Refresh the page
* See if there are any errors

Yes, 2 errors. See screenshot:

more info to come shortly ...

[chrizilla]

browser.storage.local.get().then(console.log).catch(console.error);

Thanks.
The object logged to console has 59 keys:

• 50 keys contain my 50 missing scripts
• I was able to rescue 49 of the 50 scripts
• 1 of the 50 keys with the name of one of my 50 missing scripts has value: "undefined"

This is the corruption in the local storage.

Devtools shows the same:
Firefox DevTools > Storage > Extension Storage > moz-extension://{{UUID}} > key #14 > value > undefined

[erosman]

You can delete that key (right click and select delete that key) and try again.

It is easier to:

• Go to about:debugging#/runtime/this-firefox
• Under FireMonkey click Inspect
• Go to: Storage tab -> Extension Storage

[chrizilla]

With your help, I was now able to rescue 49 out of the 50 missing scripts. Thank you so much !! ❤️ 🙏
For the corrupted script (key #14 with value undefined) I fortunately have another backup. So I am all set now, many thanks.

It would be interesting to find out, what caused the corruption in key #14.

I see 2 main possibilities: Either the DB got corrupted ...

• (a) ... when I entered the lint rules in the CodeMirror options window.
• (b) ... or when I last saved the script missing from key #14

indicators and counter-indicators:

• indicator for (a): recency
  This was the last action I took. I entered the lint rules in FM,restarted Firefox immediately and the scripts were all gone from FM.

• counter-indicator for (a): location
  The lint rules I entered survived the corruption. I can still see them in FM options as well as in key cmOptoins.

• indicator for (b): location of corruption
  key #14 is the only location I see affected by the corruption

• counter-indicator for (b): recency
  I last worked on the script stored in key #14 several days ago. But I don't know when I initiated the current Firefox browser session. It may or may not predate my work in key #14. If my Firefox session is older, it may well be that the corruption in key #14 occurred when the script was last saved to key #14 a few days ago, but I only saw the effect of this corruption yesterday, when I restarted Firefox. But then, how was FM capable of saving / loading other scripts successfully after the corruption occurred ?

As I said, I have been able to recover all lost data (at least as far as I can see from a quick initial inspection). So for me, the chapter is closed. But I am of course willing to help if you want to investigate a possible bug. I am also willing to share the corrupt files if you want them.

[chrizilla]

@erosman : I will backup the corrupted Firefox profile folder, so you can get everything from there later on.
But if you want anything from this live Firefox session, please tell me soon, because I need to continue my work and therefore reinstall FM.

[erosman]

The issue could be the rule.
It can also be closing the Options page too quickly after a SAVE which may interrupt the process due to its asynchronous nature.

What did you enter for the rule?
In FM2 it should have been:

{
  "jshint": {
    "curly": false
  }
}

In any case, it is recommended to make regular back-ups of preferences (Options -> export).

[chrizilla]

It can also be closing the Options page too quickly after a SAVE which may interrupt the process due to its asynchronous nature.

I never closed the Firemonkey tab quickly after saving. But what I may have done is saving a script and then selecting a different script (too early?). If this is somehow dangerous, it would be good to add a safeguard in the code, if possible.

The issue could be the rule.
What did you enter for the rule?

I used the wrong rule syntax.
Here is the key/value pair as stored in Extension Storage:
cmOptions:"{"curly": ["warn","multi"]}"
But do you think simply using wrong syntax can corrupt the entire storage database ? If yes, a safeguard should be added here.

[erosman]

I never closed the Firemonkey tab quickly after saving. But what I may have done is saving a script and then selecting a different script (too early?). If this is somehow dangerous, it would be good to add a safeguard in the code, if possible.

It should not happen when switching scripts. I do that very regularly and rapidly (that was the reason I designed FM to show script list on the left pane for quick access).

Even closing the Options page quickly, should not cause any problems. I have never experienced any issues myself.
The only time I have heard of problems is when the HD is full or there are RAM issues, but that has nothing to do with the extension.

But do you think simply using wrong syntax can corrupt the entire storage database ? If yes, a safeguard should be added here.

It should have rejected the rule. However, I have been working on FM3 for the last year and not used FM2.

[chrizilla]

In case that helps:
I have inspected the corrupted data blob in the .slite file in which the missing/corrupted script is stored.
The blob has a length of 18 bytes instead of the ~36000 bytes it should have:

10 3C 03 00 00 00 00 00 F1 FF 00 00 00 00 01 00 FF FF

The data is cut off after these 18 bytes.

[chrizilla]

Firefox DevTools displays the 59 keys (50 of which are my script names) in correctly readable plain-text.
But external sqlite readers operating on the .sqlite file display the 59 key names with a -1 Caesar/ROT shift.
Is this

• deliberate obfuscation applied by Firefox (independently of Firemonkey) ?
• deliberate obfuscation applied by Firemonkey itself ?
• or a result of the database corruption ? (unlikely)

[erosman]

deliberate obfuscation applied by Firefox (independently of Firemonkey) ?

FireMonkey saves the data as is, without any change.