epubcheck, any software can pass its validiation check?

[dmapr]

Originally Posted by pdurrant

I think that epubcheck writes error reports out to stderr rather than stdout, which would explain why you're only seeing the version statement.

That's right. Fortunately, the remedy is simple. Instead of

java -jar epubcheck-1.0.5.jar colomba.epub > verifyout.txt

one needs to say

java -jar epubcheck-1.0.5.jar colomba.epub > verifyout.txt 2>&1

My thanks to all three of you that have responded to my query so far.

Dmapr: Yes, once I knew what the output was being directed to I found that solution. So that problem solved.

The file name I am using, mimetype, and its content, application/epub+zip, are correct so it must have something to do with how I am doing the compression. I have always used 7-Zip File Manager according to the following steps:

add mimetype to create the archive using Compression Level – Store

add OEBPS folder (containing all book content files) to archive using Compression Level – Normal

add META-INF folder (holds container.xml) to archive using Compression Level – Normal

Change file name to epub extension (from zip)

This has always produced an epub file readable as far as I am aware by all devices; personally verified Sony 900, Sony 505, Calibre, Firefox epub plugin, and Adobe reader.
This is why I have never worried over that message before, but if there is another suggestion for how to do the compression to produce a completely 'legit' file I have open eyes.

[pdurrant]

Originally Posted by Hamlet53

This has always produced an epub file readable as far as I am aware by all devices; personally verified Sony 900, Sony 505, Calibre, Firefox epub plugin, and Adobe reader.
This is why I have never worried over that message before, but if there is another suggestion for how to do the compression to produce a completely 'legit' file I have open eyes.

Look at your epub with a hex editor. If bytes 0x1E to 0x39 are not mimetypeapplication/epub+zip then it's not a valid ePub file.

[DaleDe]

Originally Posted by pdurrant

Look at your epub with a hex editor. If bytes 0x1E to 0x39 are not mimetypeapplication/epub+zip then it's not a valid ePub file.

Even with something like notepad the text above will show up clear on the first line. That is why it is first and not compressed.

Dale

So while I would have loved to be a writer, that was not my Muse. I became a Chemical Engineer instead [BS & PhD at Cal, go Bears] and had to make do with no Muse at all. I did learn the scientific experimental method though and thought I would apply it.

Someone recently uploaded an illustrated version of Kim by Kipling here that seemed likely to pass epubcheck.

Step 1. Download that epub file an run epubcheck on it. As expected no errors or warnings.

Step 2. Decompress this file using 7-Zip and then compress that back into an epub file using the procedure described in my previous post.

Step 3. Run epubcheck on the file from step 2. Result:

 ERROR: Kim2.epub: length of first filename in archive must be 8, but was 9

Check finished with warnings or errors!

So that's interesting. How do others do the compression, with what program?

Note changed file names to Kim.epub & Kim2.epub

[DaleDe]

I use info zip from a command line or just let windows do it using drag and drop. I use 7zip for other things but not this.

[JSWolf]

I use WinRAR.

[dmapr]

Originally Posted by JSWolf

I use WinRAR.

+1. Same here.

[Chang]

Originally Posted by Chang

Maybe this goes under this thread..

I was wondering why my ePubs pass the epubcheck 1.0.5 validation but with some books ADE gives a warning "The document appears to have minor errors that might cause it to be displayed incorrectly". This doesn't even happen every time with same ePub, only sometimes.

Does someone know what's the logic behind ADE's warning because afaik it uses epubcheck validator also or am I totally wrong?

Originally Posted by pdurrant

ADE does it's own validation in addition or possibly instead of using epubcheck.

Would be nice to know does ADE know something more than epubcheck. Should I be worried about the warning message?

[ShawnM]

Originally Posted by kovidgoyal

Just so you know: epubcheck is totally meaningless. A file that *passes* epubcheck may or may not work with any given epub renderer. A file that *does not pass* epubcheck may or may not work with any given epub renderer.

I agree with you.

This is what I find strange...I got two different results running two different checkers. These were epub books that I produced for an author.

i) I ran ePub check (version 1.2) and it said there were some errors.
ii) I used Atlantis' epub checker and it passed with no errors.

Does epub check have some flaws I should know about?

The books render fine on every ebook device, so I know there are no "known issues", or formatting errors, that I need to worry about.

[ATimson]

The only known issues that I'm aware of involve OEPBS 1.2 syntax OPF files not validating because it can't find the unique-identifier. But you can look at the official bug list yourself.

[Jellby]

If epubcheck says it has some errors, the file is most probably not valid. That doesn't mean it won't work in any reader, it might, or it might not, but the file doest not comply with the specs.

[Toxaris]

I rely on FlightCrew only...

[wannabee]

I use both so when one gives an error in martian maybe the other will give you a clue of what the heck it was talking about.

[charleski]

Originally Posted by Hamlet53

First I am running under Windows XP and would like to have the epubcheck output write to a file so that I have the written output and also can use a BAT file to run it instead of typing the whole command at a command prompt each time. The trouble is that a BAT file of:

java -jar epubcheck-1.0.5.jar colomba.epub > verifyout.txt

produces the file verifyout.txt with content consisting only of:

Epubcheck Version 1.0.5

Use

java -jar epubcheck-1.0.5.jar colomba.epub 2> verifyout.txt

to get the stderr text.