|
Sigil-0.8.900 released for testing - Wait for Sigil-0.8.901
Hi All,
Just to let you know that we have made our first release of the new version of Sigil (version 0.8.900). This is the first of the pre-release series for Sigil-0.9.0. For complete details see our wordpress site: http://sigil-ebook.com To download binaries for Mac OS X and Windows, please see our github releases page: https://github.com/Sigil-Ebook/Sigil...es/tag/0.8.900 Please note, the building from source instructions have changed greatly, so we have included new instructions for building on Mac OS X, and Linux. Also, please run sha256 sum on your downloads and compare them against those posted in the release on github to make sure you have downloaded the exact same binary that we have uploaded. Please post any bug reports or feedback on this release in the Sigil forum and we will try our best to address any serious issues in quick follow-up release if needed. Thank you for using Sigil and helping to make it better. KevinH and DiapDealer |
1 Attachment(s)
W10 x64 error when I tried to use
|
Quote:
Thanks for your effort in keeping Sigil alive and making it better everyday! |
Hi theducks,
Quote:
The quickest solution is to properly download and install the very latest MS Visual C++ Redistributable Packages for VisualStudio 2010 *and* VisualStudio 2013. I recommend going only to the official Microsoft site to download these missing package. Perhaps this link may help, it came up when I did a google search and seems to be an official Microsoft link: https://support.microsoft.com/en-us/kb/2977003 Thanks for the head's up. KevinH |
Quote:
If you could; please try to download/install the Visual C++ 2010 Redistributable package for your OS (make sure it your 64/32 "bit"-ness), and see if that fixes things (I would be extremely surprised if it didn't). If so, I may actually have to distribute both the 2010 and the 2013 redistributables with Sigil (at least until the next natural Convergence of Python3, Sigil, and Qt5's VC Runtimes occurs (currently running in 180 year cycles I believe). ACK! Sometimes I wonder why I bother trying to make things work on Windows. EDIT: I'm a slow typer. ;) Shouldn't need to download/install the VC 2013 Redistributable; as Sigil already installs this. |
First of all, as Pablo, I want to thank you Kevin and Diap for all the work that has been done since version 8.6: it's really great, especially when you consider the problems to solve associated with all the internal changes and different operating systems.
And speaking of that, there seems to be some problems (Win 7 Home Premium 64b) with the Table of contents creation: most of the time nothing happens, in other cases Sigil stops. I will use the software a little longer and see what it is. |
:thanks: for all your hard work. An impressive list of changes! :thumbsup:
And duly flagged as out-of-date on ArchLinux. Hope to see how it looks RSN. |
Quote:
There's just so many features that Sigil has that I'm afraid that--speaking for myself--I fall into the trap of only testing the features I use (or the ones that I can remember actually exist!). So if steps to consistently recreate an issue can be determined/shared, that would be fantastic. |
Hi Arios
When it happens again, could you please post a copy of the ebook you were trying to build a ToC for (privately if need be) and pm DiapDealer and/or KevinH with the link so that we can recreate the problem and get it tracked down. Thanks, KevinH Quote:
|
Hi eschwartz,
Would you by any chance have access to an Arch Linux machine that runs Plasma 5. We have a bug report that all versions of Sigil (all current and all earlier) are segfaulting on Arch Linux ONLY when running the Plasma 5 Desktop (with Qt5.5) but running okay if any other Desktop is used. I have nothing to try and track that down with. I am guessing it is a conflict between Qt 5.5 and Plasma 5 that Sigil somehow triggers. Thanks, KevinH Quote:
|
here's funny bug. (windows 7/64 bit) put:
<p><a href="http://www.google.com"><mbp:nu>Google</mbp:nu></a></p> at the start of empty ebook. press save. and again. and again. :) pretty print gumbo does not work very well :) solution is easy. don't clean and format file on save or open :) |
Hi,
PrettyPrintGumbo normally works well for actual html5 tags and most others. That said, if given an unknown tag it will assume it is block tag and block tags always have carriage returns after them but in this case a block tag is occuring inside of a inline tag (the anchor) so it is confusing the prettyprint code. I will look to see how to prevent multiple carriage returns from being added with each save when an unknown block tag is added inside of an inline tag. Thanks for the head's up and test case! KevinH Quote:
|
Quote:
|
Quote:
That's something I was meaning to mention anyway. I've found the Gumbo Parser option to be far and away the least intrusive of the two options. Even with not-well-formed code or unknown tags, it does a fantastic job of leaving the code unmolested (for those of us--myself included sometimes--who are obsessed with the "aesthetics" of our code ;) ). With Tidy, we had two choices. Which in all reality were: 1) "Mess with My Code a Lot" and 2) "Mess with My Code a Lot While Adding a Lot of Extra Whitespace." ;) Now we truly have a "Try Very Hard Not to Do Anything at All" option. And that is the second of the two, "Gumbo Parser." Though we provide the Gumbo Pretty-Print option (and will continue to tweak it as issues arise), it is our hope that those who have specific needs with regard to the "prettiness" of their (x)html code might consider switching to the Gumbo Parser only option and write a Pretty-Print editor plugin to meet all their coding feng shui obsse--err... I mean requirements. :) EDIT FOR CLARITY: The two options are called "Gumbo Pretty-Print" and "Google Gumbo-Parser." It's the latter of the two that I consider to be the least intrusive clean-source option. |
I got it working
I installed the wrong (:o 32 bit) version of the redistributable the first time. |
Quote:
|
Quote:
Quote:
|
Hello:
8.9 is a pretty solid release. I’ve tested it on about 10 different books and have only run into one problem so far: the add cover tool does not work as expected. After adding or replacing cover.jpeg, invoking the tool just hangs indefinitely with the Windows “circle working” prompt. This is relatively minor stuff for such a big under the hood conversion. Great Job!! I’ve tested on my Vista Home Premium Machine with SP2 & all Microsoft KBs installed and will also try to duplicate the glitch on Windows 7 when I get the chance. More testing to come… Pete |
nother fail
1 Attachment(s)
Using the (NCX) TOC tool :rolleyes:
Also A fresh Calibre converted to EPEB causes Sigil to pop a not well formed message. (repeateable on another file) |
Hi,
On Mac OSX the Creae toc (both html and ncx) seems to work just fine for my test cases. So please post a link to the problem epubs so that we can recreate the error. If you need to you can pm either Diapdealer or KevinH with the link if the epub is private. Thanks for testing and reporting back! KevinH |
Hi,
Found the bug. We grabbed a writelock and then tried to grab a readlock when we already had the writelock, so it hung in a deadlock. I just committed a fix for this to Sigil master. It will be part of a hopefully quick future update. Thanks for you bug report. KevinH Quote:
|
Hi,
When I get back from work I will dig it up and post it o you. Thanks for your help! KevinH Quote:
|
Hi najgori,
Thanks for your test case and bug report. I have just now committed a fix to this to Sigil master which will appear in the follow-up release (hopefully soon if we can get some testcases and these other bugs tracked down). Thanks, KevinH Quote:
|
Quote:
|
Quote:
|
Quote:
|
Hi Pete6055,
I have tried using the Generate TOC using the button on the tool bar with 10 different epub test cases and in all cases it worked just like it was supposed to. But this is on Mac OS X. So would you please post a specific test case you know will create the crash on Windows. I will fire up my Windows VM and try to see if I can get the same epub to crash. Thanks, KevinH ps. I fired up my Windows 7 32 bit VM and downloaded Sigil-0.8.900 for Windows and installed it and ran it over the same 10 books using the Generate TOC icon and never had an issue. So we will really need a test case to track this one down. Quote:
|
Same here. I'm able to generate the ToC from Headers on Windows 7 with no crash. If it's repeatable, I think it's time someone just gave us one of these epubs it's happening with. Might save save some time. I'll also put 0.8.900 on my Vista machine and see what happens.
|
Quote:
EDIT (by DiapDealer): copyrighted ebook attachment deleted. Please don't post copyrighted material here at MobileRead. Contact either Kevin or myself through PM if you have a test case for Sigil debugging |
The good news is ... I can get Sigil to crash when generating an NCX from headers on Windows Vista. Just about any epub will do it for me. The bad news is ... I have no easy way of doing any real debugging on this tired old heap it's happening on. :(
|
Hi DiapDealer,
You might want to try updating to the latest VS C++ Redistributables for VS 2010 on that old vista machine to see if it helps or not. I also think there is an old VS debugger you can run to at least get some sort of backtrace. I will look to see if I can find it on my Windows VM. KevinH |
It's sigilgumbo.dll that's faulting. Check your email Kevin.
@Pete6055. If you could verify that it's the same faulting module on Windows 7 (sigilgumbo.dll), by checking the Windows event log after it crashes, that would be great. |
W10 x64
TOC from Headers crashes Split file crashes |
Quote:
|
1 Attachment(s)
Quote:
|
Quote:
Spoiler:
|
Quote:
|
Hi,
I think I have a possible fix for this error. It may be happening only on Windows because of how QString toUtf8() returns a QByteArray whose address is actually a pointer to the utf8 string of characters on Linux and Mac but not on Windows. So instead I should be using toUtf8().constData() to get a proper pointer for all platforms. Each time I took that incorrect shortcut it was in the gumbo_attribute_set_value and other gumbo routines which maps exactly to the symbol table of where the segfaults are happening. I have pushed the fix and hopefully tomorrow after work DiapDealer will be able to tell us if it was successful. Thanks for your bug reports. Until we get this nailed down please don't use Windows Sigil-0.8.900 for any type of production "just to be safe". Thanks! Kevin |
Wow guys, great work! It is always hard to make such changes under the hood. And your bugfixing speed is great as well!
|
Hi,
On a related note, is there some place where all of the <mbp:*> style tags are given in a list with something to indicate if they are block or inline types of tags. The right way to handle prettyprint is to actually include these tags (since they will be seen in ebooks) into how prettyprint classifies inline vs structural vs block vs other tags. Any pointer to a reference with a complete list of mbp tags would be appreciated. Thanks, KevinH |
| All times are GMT -4. The time now is 05:58 PM. |
Powered by: vBulletin
Copyright ©2000 - 3.8.5, Jelsoft Enterprises Ltd.
MobileRead.com is a privately owned, operated and funded community.