-updatebatch now will (or at least should):
*Download new batch and name it [updater]*.bat
*Open that script in a new CMD window.
*Exit
The [updated]*.bat script should:
*Copy itself overwriting the original batch (without renaming).
*Start that script in a new CMD instance.
*Exit.
The new script, with the original name, should:
*Delete the [updated]*.bat script
*Begin the normal script routine.
@earthing do you think I should still rename the scripts to .old or something before overwriting/deleting?
It ended up being a mixture of the previous commit and the fix. It writes a temporary file on the go that only holds preferences, and generates the target file at once at the end. It's slower than before, but it works.
While I figure out a fix for the missing characters...
Enclosing the whole merging loop in parentheses and replacing the source file with the entire output at once is more efficient than appending individual lines with >>%~2. The script doesn't have to wait for the HD to continue processing.
Everything in a line after a powershell call is considered as being called from PowerShell.
>nul didn't work because of that. Enclosing the line in brackets should fix it.
To account for the possibility of the user running the script silently in the background. PAUSE would leave an instance in memory doing nothing indefinitely.
I was going to use TIMEOUT but PING performs better.
- keeps all user.js.parrot lines intact
- keeps empty lines intact
- fix for keeping `!` and `^` in non-"user_pref" lines intact
+ some other minor changes + streamlining
You had it right the first time earthlng. Eg Start commits for 55-beta date shown is 9-July. 55-alpha release is dated 18-Aug and we drop the "-beta" part (look inside the release downloads). Start commits for 56-beta date shown is 12-Sept. 56-alpha release is dated 2-Oct and we drop the "-beta" part. And because you created the 57-alpha release before you reversed the date+version, that too is all good.
Controls the visibility of the "Options>Privacy & Security>Site Data" section.
I'd prefer to remove this completely because it only adds to the confusion about all the different storage types.
This is just an extension for localStorage (2705) with 3 methods: estimate(), persist() and persisted(). A site can ask for permission (?) to persist data which when granted basically just means that "Storage will not be cleared except by explicit user action" whereas otherwise when not persisted "Storage may be cleared by the UA under storage pressure." - I don't see a problem with that.
We'll keep 2706 inactive for now but might remove it in a future commit.
https://github.com/ghacksuserjs/ghacks-user.js/issues/264#issuecomment-345462158
- It can now handle read-only files.
- it is somewhat more explicit regarding what it's doing in some circumstances. For example, it now informs the user when no changes are made.
- It now accepts two parameters: `-unattended` and `-log`
- Minor improvements here and there.
Sorry, but AFAIK, with this enabled it clears web extension storage when clear "offsite website data" is checked on close or manually (which we do in the user.js). Note also that even with this enabled, the UI settings are disabled, and the data-on-disk calculation never finishes, so at this point, its a bit useless to enable it until we figure that out. Will be back in 7 days
this should now work no matter how the script is called (including symlinks) on both Mac and Linux.
+ Storing and restoring the original working directory to prevent problems in certain circumstances.
