Sigil-0.9.4 Released

[rubeus]

Hmmm none of my books does have a marc:relators and i've never seen this in an epub2 before and i often used the dialog to change the author as file-as isnt really followed by a lot of readers so i always have the author given as lastname, filename
Doing this today with 0.9.4 i'm getting now this unwanted entry.

Just to make it clear: i didnt added marc_releators, it was added automagically.

[DiapDealer]

For the record Kevin: both Flightcrew and Epubcheck claim that the attribute "opf:scheme" is not allowed in the creator element. Same for the contributor element.

This is what the editor produces now:

<dc:creator opf:file-as="Hope, Anthony" opf:role="aut" opf:scheme="marc:relators">Anthony Hope</dc:creator>

Older versions of the Metadata Editor produced the same thing--but without the opf:scheme="marc:relators" attribute/value.

[KevinH]

Doitsu said the same thing. Under epub "2.0" spec of 2007, you need the opf namespace to use the file-as, scheme, and role attributes as they are not part of the dc namespace nor allowed for those metadata elements. So what is here should be technically correct.

Doitsu pointed out that under epub "2.0.1" spec in 2010 file-as, scheme, and role have been added as allowed attributes for the dc: namespace metadata tags. He said that epubcheck uses the "2.0.1" spec which means they do not need the attribute namespace additions (none of them).

Once I get back I will look at this, and if this is incorrect, we can release a quick Sigil-0.9.5 to fix this, the wcc validator, and the bug in importing HTML metadata

KevinH

[KevinH]

Had a free moment when getting gas ... in epub "2.01" opf:scheme does exist but only for dc:identifier and not for dc:creator or dc:contrinutor as the opf:role implicitly implies the marc:relators scheme as the only scheme allowed for opf:role. That makes no sense at all but it is the epub 2.0.1 spec. It is the exact opposite for what they did for epub3 which requires the scheme used for role refines. So I will fix this when I get back and we can do a quick bugfix release of 0.9.5 since 0.9.4 has two regressions now and added a feature to allow better css validation choice.

[theducks]

Sigil detects EPUB3
I have an ancient RMSDK device. KISS, EPUB2 are best

But now it won't let me downgrade (easily) those to EPUB2.
I don't want a NAV file at the beginning (I already move the inline TOC to the end, if it exists, and if contains more meaningful entries otherwise: DEL)

[KevinH]

In general downgrading an epub3 to an epub2 may not be easy given the epub3 version may use features not available under epub2, such as epub3 extended metadata, scripts, html5 specific tags, epub3 specific attributes, and etc. So simply allowing you to change from 3 to 2 inside Sigil means the ebook would be in some sort of intermediate state preventing Sigil epub version specific tools or menus from working. So that is not going to happen.

That said, we can write an epub_3to2_plugin that will use the nav to create the ncx and guide, strip out the nav, strip out epub:type attributes, replaces section tags with divs, downconvert epub3 metadata to 2, etc. The user would have to search for other html5 specific tags and replace them with some epub2 code equivalents, place comments around or any scripts, or delete them, etc, etc.

Since calibre's editor isn't epub version specific, you may be better off handling the downgrade to epub2 using calibre until we get such a plugin written.

KevinH

[DiapDealer]

Originally Posted by theducks

But now it won't let me downgrade (easily) those to EPUB2.

It is unfortunate. I used to do the same thing with some "simple" commercial epub3s. But to keep both epub2 and epub3 editing possible with the same gui/same tools, sacrifices had to made. Changing the package version "on the fly" was one of those sacrifices.

Luckily, I've found that with those simple epub3s, changing the package version number from "3.0" to "2.0" with an external editor is enough for Sigil to open them as epub2's. Then I can continue downgrading the rest of the stuff as usual. Using 7Zip, I don't even have to manually unzip the archive. I just open the epub in the 7Zip program, edit the opf file, save, close, and re-open in Sigil.

[KevinH]

That would work too!

[theducks]

Originally Posted by DiapDealer

Luckily, I've found that with those simple epub3s, changing the package version number from "3.0" to "2.0" with an external editor is enough for Sigil to open them as epub2's. Then I can continue downgrading the rest of the stuff as usual. Using 7Zip, I don't even have to manually unzip the archive. I just open the epub in the 7Zip program, edit the opf file, save, close, and re-open in Sigil.

I can do that with EPUB Unpack (the free utility from the folk at AWP) ot Calibre, since that is my usual Launch point (open with PI) for Sigil

[KevinH]

Hi sdjskr

I spent some time trying to track this down. I simply can not recreate this at all. When you hit return the file is getting a simple carriage return linefeed pair as expected. No strange chars are being generated.

It could be a lost focus issue. So when the bug happens in BookView, please try clicking the "<>" icon to switch to CodeView and then back? Does the cursor come back?

If not, please try clicking the "<>" icon again to switch to CodeView and then make some small meaningless change in CodeView and then back to BookView? Does this make the cursor return?

If not caused by the widget losing focus, I did find some Qt code for QWebPage that had the following method:

void QWebPagePrivate::inputMethodEvent(QInputMethodEvent *ev)

that shows when an QinputMethodEvent has a cursor attribute with 0 length it can actually make the cursor invisible:

case QInputMethodEvent::Cursor: {
            frame->selection()->setCaretVisible(a.length); //if length is 0 cursor is invisible

So if not focus related it appears that somehow your input method is generating events for the cursor with 0 length attributes that are making the cursor invisible. I am just not sure why. Your video shows that the cursor position is still kept, and works but that the cursor itself was simply made invisible.

Since it appears that other Windows users can not recreate your bug either, my guess it is some conflict between other Windows system software you are using and the Qt library that comes with Sigil.

Any bug would be inside Qt and not something Sigil could fix. That said, we may be able to intercept and filter out that event if we can figure out what is generating it.

KevinH

Originally Posted by sdjskr

Hi, again!!!

At first, I wrote some Korean and Alphabet in a notepad to demonstrate how the cursor turns into. As of Sigil editor, I cannot see where I am typing because of the missing cursor as soon as I hit Enter. It's kind of frustrating.

https://youtu.be/yRyjTxIQ2dQ

I expect that to be solved soon! best regard!!!!

[jedi108]

Open an ePub with Sigil 0.94 (happens both on Mac and PC)
Save as another file
Link in ToC in iBooks no longer works.

Sigil seems to have moved the ToC file from outside of the Text folder into it, not sure if this has to do with anything.

Help!!!
Thanks

[KevinH]

i have no idea what iBooks is looking for or not looking for. Open any valid epub that validates with epubcheck in Sigil, then save it under another name then run it though epubcheck again. It should still validate just fine. If not, please file a bug report on our github site and include a link to a noncommercial epub that demonstrates the issue.
KevinH

[DiapDealer]

Sigil always relocates the files to suit its own internal structure, but it always updates the links in the various files. Is it only in iBooks that the links in the toc don't work?

Do you have a small uncopyrighted sample that you can post that will show broken links in the ToC when following your steps? So far none of my tocs have broken links when I follow your steps. Granted: I don't have iBooks to test with, but if the Epub is valid and has a working toc before opening it in iBooks, then clearly iBooks is doing something weird. For it to be a Sigil bug, there would have to be something broken immediately after the Save As.

For clarity, are you useing Save As, or Save a Copy?

[BetterRed]

Originally Posted by KevinH

So if not focus related it appears that somehow your input method is generating events for the cursor with 0 length attributes that are making the cursor invisible. I am just not sure why. Your video shows that the cursor position is still kept, and works but that the cursor itself was simply made invisible.

FWIW - losing the insertion point cursor occasionally happens to me in Windows/Firefox, in fact it happened just a few minutes ago while typing a post in the calibre forums. I solve it by refreshing the tab.

But I don't think Firefox uses Qt and especially not WebKit. I assume its something I do with the keyboard. It only happens in Firefox, but AFAIK it's the only html based program I type into regularly. I guess I'm suggesting it maybe a generic hiccup between HTML based programs that receive input and Windows.

I don't use Sigils BookView.

BR

[sdjskr]

Originally Posted by KevinH

Hi sdjskr

I spent some time trying to track this down. I simply can not recreate this at all. When you hit return the file is getting a simple carriage return linefeed pair as expected. No strange chars are being generated.

It could be a lost focus issue. So when the bug happens in BookView, please try clicking the "<>" icon to switch to CodeView and then back? Does the cursor come back?

If not, please try clicking the "<>" icon again to switch to CodeView and then make some small meaningless change in CodeView and then back to BookView? Does this make the cursor return?

If not caused by the widget losing focus, I did find some Qt code for QWebPage that had the following method:

void QWebPagePrivate::inputMethodEvent(QInputMethodEvent *ev)

that shows when an QinputMethodEvent has a cursor attribute with 0 length it can actually make the cursor invisible:

case QInputMethodEvent::Cursor: {
            frame->selection()->setCaretVisible(a.length); //if length is 0 cursor is invisible

So if not focus related it appears that somehow your input method is generating events for the cursor with 0 length attributes that are making the cursor invisible. I am just not sure why. Your video shows that the cursor position is still kept, and works but that the cursor itself was simply made invisible.

Since it appears that other Windows users can not recreate your bug either, my guess it is some conflict between other Windows system software you are using and the Qt library that comes with Sigil.

Any bug would be inside Qt and not something Sigil could fix. That said, we may be able to intercept and filter out that event if we can figure out what is generating it.

KevinH

Hi, KevinH. Thank you for the study to nail this issue down.

It could be a lost focus issue. So when the bug happens in BookView, please try clicking the "<>" icon to switch to CodeView and then back? Does the cursor come back?

As you can see in this video, the cursor comes back after switching between two VIEWs. But, it goes away in BookView as soon as I press Space bar or Enter while typing Korean.

And I agree with your point that the cursor still exists, but not showing or losing focus, because I can still type Korean without the cursor.

My guess is that the issue is somewhat related to QWebView that is deprecated in Qt 5.6 now. I know Qt 5.4's QWebView was used to create Sigil. I've found some people having discussion of the topic. They are saying the disappearing caret, commonly occurred to one who is using QWebView. They are also talking about Sigil.

link

Which means, the cursor problem is not my own issue, some other Korean using Sigil is having the same problem.

Anyway, I'm certain there must be some method to work around. I can try your newly compiled version applied with your idea for Windows 10 64bit on my machine. You can just send it to me at bc399@daum.net.

And I am expecting some refreshed manual to compile the source code from Sigil in Windows environment. Someone promised it. I also want to find some clue to resolve this issue.