Beta Testers for Sigil-2.3.0 Needed

[KevinH]

Hi All,

For our next release of Sigil we have made some major changes to the codebase so we are looking for Beta testers to heavily test things to make sure all works as expected:

These changes include:

1. Move to Qt 6.7.2

2. Move to Python 3.11.9

3. Wholesale changes to Sigil source code to drop support for older Qt5

4. Reverted a large number of late changes that impacted all buttons and checkBoxes that made users require more clicks to get work done. This should bring us back towards Sigil 2.1.0 in that regard but hopefully not hurt tab to focus keyboard users.

5. All bug fixes we have accumulated since the 2.2.1 release

6. Redesigned changes to use Unicode NFC normalization form so that all epubs built with Sigil can be searched equally well in all languages and all good e-readers as well as with Sigil itself. This is to address concerns of many epub developers who make heavy use of accented languages and those that work with RTL languages like Arabic and Hebrew.

Both DiapDealer and myself are using these builds on a daily basis and all appears to work for us just fine but because people have different work patterns getting feedback from a larger group of testers is important.

So I have posted CI builds (Beta) for Sigil 2.3.0 in my personal repo on github here:

https://github.com/kevinhendricks/Bu...nMac/releases/

The 3 files available are:

Sigil.app-230Beta-Mac-x86_64.tar.xz
Sigil.app-230Beta-Mac-arm64.tar.xz

and for Windows:

Sigil-installer-2.30Beta-Windows.zip

Note: Windows users will need to unzip the download to get to the normal setup installer.

So please if you have any free time and interest in helping Sigil, please grab and use any of these Beta versions and report back any results, good or bad with your normal epub workflows in this thread, so we can track issues down and get them fixed.

Thank you!

Your Sigil Developer Team

[BetterRed]

Windows 10 keyboard centric user

No immediately obvious differences, for me everything looks and works the same. The only vague expectation I had from recent issues raised here was that the first item in list boxes would be selected when tools such as the ToC Editor and Spell checker were first used.

So this rather than this

Doesn't bother me that its not selected, tap space 'fixes' it.

BR

[KevinH]

The EditToC dialog was what the bug report was about. The list box itself now has focus but nothing is selected. This makes sense as technically this list could be empty. Using the down arrow to actually select a line works in my testing. No tab was needed to shift focus anyplace. I assume your space just selects the first line like a down arrow would.
Since when editing a ToC there is no rule you must start at the top. Giving the List widget focus is the right thing to do here.

I did not see a report about SpellCheck but I will look into that. Thanks.

Update: Checked out the code for Spellcheck and there are no changes there from earlier Sigil 2.1.0, so click or tab to focus on what you want will still be required.

[BetterRed]

The same "nothing selected" behaviour I saw in the Spellchecker, is in the ToC Editor and the Metadata Editor - as can been seen below:

This rather than

This rather than

That suggests to me that it is a behaviour inherited by those dialogues from underlying object classes - presumably from the Qt library.

In the ToC Editor, if I don't hit space to select the "Title" line entry then… when I tab to the Down arrow button and hit space the "Title" line entry doesn't move down… because nothing is selected.

Added: Given that's how it's always been I DO NOT regard this as a bug.

Other dialogues containing lists are the same… e.g Reports, Keyboard shortcuts.

I suspect I misread the discussion about the ToC Editor buttons.

BR

[KevinH]

Prior to this version, you needed to tab or mouse click to get focus in the List or Tree at all. Now it has focus but nothing is selected until you use an arrow key to select a cell or line. And previously using a button stole focus away from the List or Tree but that should no longer be happening.

[DiapDealer]

Wanted to give folks an example of the inherent Qt Dark mode available on Windows by unchecking the "Use Classic Sigil Dark Theme" in Sigil preferences. I'm using the green Material icons with Windows' own Teal accent/highlight color. I'm adding just a touch of custom styling via the qt_styles.qss file:

QToolBar{
   border: 0;
}

QTabBar {
  qproperty-drawBase:0;
}

I have no intentions of eliminating the original classic Windows dark theme (so long as it it doesn't become impossible to support both), but the inherent Qt dark theme may become the default on new Windows Sigil installs in the future. Hope to post some Linux themed versions of Sigil soon.

[oston]

I've been using V 2.3 with my normal workflow for the past hour or so and have not encountered any issues. I'll have more time tomorrow to explore other things.
Jim

[DNSB]

I ran into an issue after downloading the Windows installer .zip file. Microsoft Defender complained that the .zip file contained Tojan:Script/Wacatac.B!ml and deleted the file. I redownloaded it and used 7-Zip to extract the installer. While MD complained about the trojan, it did not delete the file until after 7-Zip closed. I ran the .zip file through a couple of online multi-scanner sites and no complaints so I'm going to call it a false positive.

[DiapDealer]

I hate anti-malware software. I really do.

I got no (updated) Defender prompts/warnings when using the above links to download/install. Using 7-Zip as well.

[KevinH]

That Windows installer is a straight download from our github actions CI build onto my mac then drag and dropped onto my repo.. For there to be a trojan of any kind, it would have to exists inside our github actions build sequences.

That is not impossible but highly unlikely. Are other Windows users seeing similar warnings?

Are false positives by Microsoft Defender like this common?

[KevinH]

Does MD use "AI" heuristics? Do they look at how many downloads have come from my repo, how many users have downloaded it, look inside and see a exe and then freak! If so, that kind of shit should be banned. Either it is infected or it is not but do not say it is based on predictive "guessing" instead they should use of an accurate binary signature.

[DiapDealer]

We might try posting a link to the unzipped installer and see if it makes a difference.

For distributing betas we're probably out of luck with this sort of thing, but for Windows users who get warnings on official Sigil releases and are uncomfortable exempting Sigil and its folders, the best option is to install Sigil via the winget or Chocolatey repositories. They vet their submissions pretty thoroughly before making them available, and Windows seems to trust the software installed with their clients.

Chocolatey's verification page for their Sigil 2.2.1 installer (downloaded directly from our github release page): https://www.virustotal.com/gui/file/...0de-1719090708

[KevinH]

This evening I will unzip and post the exe alongside the zip just to see what MD makesof that.

[KevinH]

Okay, had few minutes. Unzipped that installer zip on my mac using unzip and then uploading the resulting file to my repo:

https://github.com/kevinhendricks/Bu...nMac/releases/

Grab:

Sigil-2.3.0-Windows-x64-Setup.exe

And let me know if that in any way impacts how MD reports the file.

[DiapDealer]

Thanks!
I've had no Defender issues with either download. But then none of my machines are strangers to downloads from your GitHub account.