0.6 Beta take 3 (0.5.906)

[crutledge]

Originally Posted by kiwidude

@crutledge - all that shows is that you have some html pages that don't have file paths in their hrefs. There is no indication of what led to this or whether they ever had file paths in the first place? As I said above we need a repro case if you think Sigil is the cause - the document you actually started with which had paths, and what steps you followed which led them to be dropped.

1. Load html to sigil.
2. Press F6
3. Create toc
4. Go to in-line toc or loi

You will find NO full paths.

Previous version of 6.0 beta worked fine.

[kiwidude]

@mrmikel - can you give an example?

[crutledge]

Originally Posted by kiwidude

@crutledge - all that shows is that you have some html pages that don't have file paths in their hrefs. There is no indication of what led to this or whether they ever had file paths in the first place? As I said above we need a repro case if you think Sigil is the cause - the document you actually started with which had paths, and what steps you followed which led them to be dropped.

1. Load html to sigil.
2. Press F6 (split)
3. Generate sigil toc
4. Go to in-line toc or loi

There are no paths.

Previous beta worked fine.

[meme]

@crutledge - seems ok to me.

1. Start Sigil
2. Add Existing file The Chouans.htm
3. Pre F6 split - does nothing in this version because the legacy marker name is used - or did you change them (as the issue might be split)
4. Generate TOC
5. Create HTML TOC

toc.ncx - full paths
TOC.html - full paths

Not sure what loi is.

[kiwidude]

@meme - I've managed to replicate the problem (not with crutledge's attachment, which as you say works fine). It is actually nothing to do with TOC at all. It is that splitting is now broken for an href to a link on the same page - after the split the href on the original page is not being updated to point to the second page.

[Doitsu]

Originally Posted by crutledge

1. Load html to sigil.
2. Press F6
3. Create toc
4. Go to in-line toc or loi

You will find NO full paths.

This might be a x64 build issue. Generating an HTML TOC via Table of Contents > Create Table Of Contents worked fine on my 32bit XP system.

For example, I got <a href="../Text/The%20Chouans.htm#">TABLE of CONTENTS</a> for one of the TOC entries in your test book.
(F6 doesn't work with your file because the marker names have been renamed from <hr class="sigilChapterBreak" /> to <hr class="sigil_split_marker" /> and Sigil never had a LOI feature.)

Try installing the 32bit version of the latest beta and see if that makes a difference.

[kiwidude]

@Doitsu - sadly in this case it is the code at fault (and me to blame). In attempting to fix something else this week I broke the splitting where the hrefs are linked to the same page. Schoolboy error.

Edit: Ok, worked out the fix for our next build. Ironically it was an attempt to fix an old Sigil issue that Crutledge reported on the last thread which did the damage, of hrefs in the form "SomePage.html#" (trailing # but no fragment) not being recognised. Methinks the next Sigil beta build will be sooner rather than later once user_none finishes slapping me around...

To summarise - it is only Split which is broken with this build and only in one circumstance:
<a href="foo"> - No problem
<a href="foo#"> - No problem
<a href="foo#bar"> - No problem
<a href="#bar"> - If after split the #bar is on a new page, will not update href

As a workaround (such if you are importing a single HTML page to split it) just do a find replace to insert the name of that page before any links, so href="#bar" becomes href="../Text/foo.html#bar" where foo is your page name.

[theducks]

Were you trying to fix the long time problem in Sigil where you end up with
href="../Text/#bar" after a split, because an anchor is no longer on the same page)?
Would it work to insert the file name (make it absolute rather than relative) upon opening a document (while it is doing the other SGC... inserts)?

[kiwidude]

@theducks - yes that particular long term issue got fixed, as did several others we found or were reported. Sadly one of the fixes through a silly assumption introduced the new bug above. A simple Find/Replace to turn href="# into href="../Text/Mypage.html# will allow the split function to work as it is supposed to until we get another build out.

[mrmikel]

Originally Posted by kiwidude

@mrmikel - can you give an example?

Here is a picture.

[kiwidude]

@mrmikel - when you said parenthesis, I thought you meant ( ) not a single quotation mark .

That spellcheck behaviour is unchanged from the previous beta. I know exactly why it will be happening though, we will take a look at it. Thanks for the feedback.

[mrmikel]

DUH--Too early in the morning. Been up for hours trying to update two computers into Windows 7, and install a virtual machine on one so I can run a DOS database program in it....single quotation mark, single quotation mark

[Moonraker]

I don't post very often in MR but I just had to say a big thank you for the latest version of Sigil.

There are so many new things to use which will help me enormously.

I am so impressed and can't believe my luck at being able to use this wonderful application.

THANK YOU!

[ludovicus]

Not working on OS X (10.6.8) either.

Console report & crash log:

Spoiler:

10/1/12 2:04:19 PM com.apple.launchd.peruser.502[106] ([0x0-0x539539].com.sigil-ebook.Sigil.app[17001]) Job appears to have crashed: Segmentation fault

Process: launchd [17015]
Path: /Users/photo/Desktop/Sigil 906.app/Contents/MacOS/Sigil
Identifier: com.sigil-ebook.Sigil.app
Version: ??? (???)
Code Type: X86-64 (Native)
Parent Process: launchd [106]

Date/Time: 2012-10-01 14:04:36.034 -0400
OS Version: Mac OS X 10.6.8 (10K549)
Report Version: 6

Exception Type: EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_INVALID_ADDRESS at 0xfffffffffffffff8
Crashed Thread: Unknown

Backtrace not available

Unknown thread crashed with X86 Thread State (64-bit):
rax: 0x0000000000000055 rbx: 0x0000000000000000 rcx: 0x0000000000000000 rdx: 0x0000000000000000
rdi: 0x0000000000000000 rsi: 0x0000000000000000 rbp: 0x0000000000000000 rsp: 0x0000000000000000
r8: 0x0000000000000000 r9: 0x0000000000000000 r10: 0x0000000000000000 r11: 0x0000000000000000
r12: 0x0000000000000000 r13: 0x0000000000000000 r14: 0x0000000000000000 r15: 0x0000000000000000
rip: 0x00007fff5fc01028 rfl: 0x0000000000010203 cr2: 0xfffffffffffffff8

Binary images description not available

[signum]

I downloaded the source for Sigil 0.5.906 beta yesterday, compiled it and exercised it a bit. I have a project of converting over 300 pulp novels from txt format into epub format and Sigil is just perfect for the task. Sigil beta exhibited a few problems you should know about. Before I start, I want to say that I'm growing to like Sigil more and more. The latest beta is a bit like Christmas morning, unwrapping the presents and then feeling sad that some broke when first used. It is not my intention at all to just pick on the flaws, but to do what I can to help get them fixed. But, you can't fix them without knowing about them. So, here's my short list of anomalies arranged in approximate order of importance to me, starting with the least important (BTW, I'm using kubuntu 12.04 “precise” 64-bit):

1) In the tools → Saved Searches screen (what used to be the Search Editor), the entities for left and right double quote are misspelled. There is no “t” in ldquo or rdquo. This happens in two places. At least it's consistent. While you're looking at this, please consider adding the entities for left and right single quote also.
2) Trying to insert a split marker at the current cursor position using the keyboard shortcut <ctrl-shift-return> doesn't appear to do anything. Using the menu item from the tool bar works just as I would expect. Also, the keyboard shortcut <ctrl-return> splits the file immediately, as it should.
3) Attempting to open an epub file that has a zero-length html file aborts (core dumps) after throwing an instance of “QtConcurrent::Unhandled exception' what(): std::exception” Normally, I wouldn't care about this, but it took some time to find the underlying condition (I wasn't expecting to have a null file). Another issue is that sometime later my system picks up the core dump and offers to send it to the (k)ubuntu maintainers. It repeats this offer every time I login until I go through all their screens from the beginning and decline everything. Perhaps you could find a way to fail gracefully with a stern error message instead of allowing it to crash.
4) Multi-select for bulk renaming doesn't work. Left click on the first file, <shift> left click for the last file in the range. The range flashes blue briefly and settles with only the last file selected. If I start with selecting the first file in the list, I have a 1-in-50 chance of being able to mark a range. Otherwise, no chance at all. Likewise, using <ctrl> click to mark a discontinuous range has never worked for me.
5) Regex search on All HTML Files sometimes won't advance to the next HTML file after single-stepping through and reaching the bottom of the current one. Other times it works, though. I've also seen it, a time or two, replace with the literal characters “\1” (no quotes). I'd be hard-pressed to reproduce either of these issues every time, though.
6) “save as” with “auto select file name extension” check box left set by some previous application freezes the window. It isn't running up cpu cycles or hitting the disk, it's just frozen as if waiting for something. Even the cursor is frozen so that I can't quit it the usual way, from the file menu. I have to go to another window using a keyboard shortcut and kill the Sigil process by its number. This situation typically arises when I use Sigil on an HTML file and try to save it as an epub. At least it doesn't core dump.