Sigil-0.9.991 (1.0 Pre-release)

[DiapDealer]

Originally Posted by DiapDealer

I see a recent commit that looks like it may have already addressed this. I'll try it out. Thanks!

And indeed it did fix it.

[KevinH]

This should now be fixed in master. The problem was letting the nav file count towards determining the default folder for the group Text. So Text had two paths, the nav.xhtml path in root with a count of 1, and a marker.xhtml path in xhtml/ with a count of 1 and the resulting sort had them tied and ended up putting the default folder as root and not where it should have been.

By manually handling the nav file, and removing it from the bookpaths to be told to FolderKeeper, this problem goes away.

Originally Posted by DiapDealer

An epub won't be very useful since it's basically already broken. But that's sort of the point. The Create a Custom Empty Epub tool is creating a broken empty EPUB3 (or at least it can).

Steps to recreate:
** Make sure your default empty EPUB version is set to EPUB3 **
1) open Sigil.
2) start the "Create a Custom Empty Epub" tool
3) add markers for OPF file, NAV File, Style Files, and Image Files directly in EpubRoot
4) add an "xhtml" folder to EpubRoot
5) add a marker for Xhtml Files in the "xhtml" folder
6) click OK (it doesn't matter if you say yes or no to the prompt to save the structure as default).
7) Double-click on nav.xhtml in the Book Browser and note the incorrect href for "Section0001.xhtml" (or run the EpubCheck plugin and note the "'Section0001.xhtml' could not be found in the EPUB" error).

[KevinH]

Great! My previous commits should have fixed both but the wrong default path for group "Text" created the problem for nav files.

Originally Posted by DiapDealer

And indeed it did fix it.

[KevinH]

By the way, any existing links in the ncx and the nav should always be automagically updated by way of calls to PerformUniversalUpdates which is done after a Move or Rename happens.

So as long as we do not have broken links, both should be auto updated.

I am thinking based on all this and based on what Doitsu reported, we should be creating a link checker function or even link validator, that will test for and report broken links that we can use before allowing renames and moves to be done.

What do you think?

Perhaps a fast link check done in python3lib?


Kevin

[KevinH]

So as long as I did not mess the Index stuff, again, I think current master has fixed all reported bugs so far.

[KevinH]

I take that back, I can recreate the issue with the nav if I use more than 1 marker.xhtml in two different directories.

I need to figure out a better way of handling template creation.

[KevinH]

Okay, I tried to improve the create empty nav and ncx code and pushed it to master. That said ... the required changes seem to make the code begin to look spaghetti-ish. So if this code requires any further changes we will have to rethink how these empty files are created and probably need to separate creation from initial text to clean things up.

[odamizu]

Thank you KevinH & DiapDealer for your insights into Sigil standard structure vs. non-standard structure. I must admit that after using Sigil, I now derive comfort in having all my epubs being standardized to the same structure — which is funny because when I first started using Sigil I was horrified that it re-structured every epub that I loaded into it!

But on to issue-reporting. I may have found one related to Page Lists, though it's not specific to 0.9.991. I thought it was, but then I went back to 0.9.18 and found it had the same issue. Not sure if this is known or expected, but I report it in case it is helpful.

• Take valid epub3 with nav and ncx, where both nav and ncx have a page list (<nav epub:type="page-list"> and <pageList>)
  
• Delete a file
  
• opf updates automatically
  
• edit TOC and <nav epub:type="toc"> updates automatically
  
• use Epub3 Tools to Generate NCX and <navMap> updates automatically
  
• however both <nav epub:type="page-list"> and <pageList> fail to update

Let me know if any additional info would be helpful.

[KevinH]

Historically, the ncx page list is simply not used for epub2 so Adobe created its page-map.xml file. Page lists are part of the spec for NCX but nothing ever seemed to support it. Same for ncx landmarks. The opf guide is used instead. Really the only thing the ncx is typically good for is the toc.

Under epub3 the nav page list should be updated for every move and rename of files, but I am not sure about deletes. I will try to recreate what you are seeing. If so it is probably a bug but one that has existed for a long time, as nav page-lists are quite rare.

Manually rebuilding the nav page-list after deletes is probably a good idea as you may want to assign that page number to some other file otherwise you will have to manually rebuild all of the page number labels!

[KevinH]

I forgot to mention that page list page numbers can not be assigned or deleted on a whim as they are supposed to match exactly the page numbers of a particular print edition of a book which is then specified in the epub3 metadata with refines. So unless the associated print book has deleted that page, it really should not be deleted in the epub version otherwise the page numbers themselves become quite worthless for academic citing reasons as they no longer match the print version they were meant to reference.

[odamizu]

It does seem this is not worth your spending time on.

The only reason I noticed it is because I got a validation error after deleting a file, and this turned out to be the cause. However, having Sigil automatically update the page list seems problematic for the reasons you cite, so it's probably enough to have the validation error alert the user that the page list needs fixing.

[CalibUser]

I used Sigil-0.9.991 with PageEdit 0.9.6. I found only one issue (that I have not been able to reproduce). My epub had 19 images in it and for some reason they were still listed in Sigil-0.9.991 but the images were not viewable in PageEdit 0.9.6 (I could only see a thumbnail that did not resemble the images). When I closed both of these programs and used Sigil 0.9.14 to open my epub I found that all my images had been removed.

[DiapDealer]

That sort of thing would suggest that the images were not properly manifested in the opf file in the first place (and removal would only happen if you saved the epub). Even earlier versions of Sigil couldn't fix wrong hrefs in the manifest when opening, but Sigil should warn you that unmanifested files are present when opening. Did you get no such warning on that particular epub?

[snarkophilus]

Looking to help out with some pre-release testing. For those who don't have a build environment set up is there any mechanism for getting the latest pre-release code, like a daily build of the master branch? I would be nice to be "ahead" of some of the fixes identified and fixed since 0.9.991 was released.

Also, if I encounter some issue with the 1.0 prerelease, will re-installed the last 0.9.18 release over the top work cleanly?

I'm on Windows 10 if that matters.

[DiapDealer]

Originally Posted by snarkophilus

Looking to help out with some pre-release testing. For those who don't have a build environment set up is there any mechanism for getting the latest pre-release code, like a daily build of the master branch? I would be nice to be "ahead" of some of the fixes identified and fixed since 0.9.991 was released.

PM sent with details.

Originally Posted by snarkophilus

Also, if I encounter some issue with the 1.0 prerelease, will re-installed the last 0.9.18 release over the top work cleanly?

I'm on Windows 10 if that matters.

Yes, it will. No problems. Might want to make a backup of your preferences folder before installing 0.9.991 that you could later restore if you wanted to be super-duper safe, but there should be no problems.