Snipped from Proposal: Extending Epub

[Nate the great]

NOTE: I moved these from another thread because I made a mistake and took the thread off topic.

Originally Posted by Peter Sorotokin

I do not think XDXF would work well because it is a single file. Russian-English dictionary is 100 meg XML file: loading that in handheld memory would be challenging. So, at a minimum, single XML needs to be broken into pieces. Also, it is not an issue how to represent the content: it can be done either by CSS-styled XDXF snippets or CSS-styled XHTML with classes. This part is OK, no changes to the standard are required. The issue is how to build an index that can quickly guide reading system to the appropriate part of the content. Note that a single index file won't cut it - it will likely be too large. Some sort of hierarchical structure broken between several files is needed. That, I think, is an extension to EPUB that needs to be defined (or borrowed).

I see the index as a feature of the reader software, not the Epub format. That's how Mobipocket Reader does it. When you look at the index of a dictionary in MobiReader, what you see on the screen was generated on the fly by the software. You're not looking at the contents of a file.

The Kindle does create index files, true. But it indexes all the ebooks on the device, not just the ones with Mobipocket's reference tags. This is a software feature that can be implemented now, without using these tags.

Besides, I thought the purpose of these tags was to remove the need for a separate index file.

[tompe]

Originally Posted by Nate the great

I see the index as a feature of the reader software, not the Epub format. That's how Mobipocket Reader does it.

Are you sure about that? I have always assumed that a MobiPocket dictionary contains an index.

[Nate the great]

Originally Posted by tompe

Are you sure about that? I have always assumed that a MobiPocket dictionary contains an index.

I'm fairly certain that it's a result of the reader software, and not part of the ebook.

[tompe]

Originally Posted by Nate the great

I'm fairly certain that it's a result of the reader software, and not part of the ebook.

I am fairly certain that my Cybook does not index my dictionary. But I am a bit unsure about what you mean by indexing.

You realize that the html tags on MobiPockets homepage is not a description of the MobiPocket format? Running mobigen the html files can be converted to anything and an index can be added to the book.

[DaleDe]

Originally Posted by tompe

I am fairly certain that my Cybook does not index my dictionary. But I am a bit unsure about what you mean by indexing.

You realize that the html tags on MobiPockets homepage is not a description of the MobiPocket format? Running mobigen the html files can be converted to anything and an index can be added to the book.

Exactly, the performance issues clearly dictate that some sort of Index be provided. Searching the dictionary as linear database would take way too long.

Dale

[Nate the great]

Originally Posted by tompe

I am fairly certain that my Cybook does not index my dictionary. But I am a bit unsure about what you mean by indexing.

You realize that the html tags on MobiPockets homepage is not a description of the MobiPocket format? Running mobigen the html files can be converted to anything and an index can be added to the book.

I'm talking about something you can't do with the Cybook. It lacks the software because it's running an early generation of Mobipocket Java code.

[DaleDe]

Originally Posted by Nate the great

I'm talking about something you can't do with the Cybook. It lacks the software because it's running an early generation of Mobipocket Java code.

Actually the Cybook runs the latest generation of Mobipocket java code and does support dictionaries. It is currently the best implementation of Mobipocket available in any eBook device with the possible exception of the Kindle.

Dale

[Nate the great]

Originally Posted by DaleDe

Actually the Cybook runs the latest generation of Mobipocket java code and does support dictionaries. It is currently the best implementation of Mobipocket available in any eBook device with the possible exception of the Kindle.

Dale

No. The Hanlin V3 runs the latest version. This has been demonstrated.

[Nate the great]

Okay. I went and built my World Fact eBook again. The log shows the that the indexes are built in to the book. I was wrong.

[DaleDe]

Originally Posted by Nate the great

No. The Hanlin V3 runs the latest version. This has been demonstrated.

the Hanlin has no dictionary support at all and its font support is quite limited. If it is the latest version then Cybook has done some serious modification. One thing I want to see in the Hanlin mobi support is to decrease the font size on superscipt items.

[Nate the great]

Originally Posted by DaleDe

the Hanlin has no dictionary support at all and its font support is quite limited. If it is the latest version then Cybook has done some serious modification. One thing I want to see in the Hanlin mobi support is to decrease the font size on superscipt items.

I hadn't known that the V3 doesn't have dictionary support. That's odd. Jinke must have screwed up when porting the software.

[Peter Sorotokin]

Originally Posted by Nate the great

I see the index as a feature of the reader software, not the Epub format. That's how Mobipocket Reader does it. When you look at the index of a dictionary in MobiReader, what you see on the screen was generated on the fly by the software. You're not looking at the contents of a file.

That's a legitimate point of view, of course. The problem is that building index on the device would be slow and drain the battery and building it elsewhere would mean that special software needs to be used to transfer the book to the device. I think that support for indexing is just too central for a dictionary to leave it out.

[DaleDe]

Originally Posted by Peter Sorotokin

That's a legitimate point of view, of course. The problem is that building index on the device would be slow and drain the battery and building it elsewhere would mean that special software needs to be used to transfer the book to the device. I think that support for indexing is just too central for a dictionary to leave it out.

I believe you are right. I suspect the easy way to generate an index is just to develop a linked list of all the words. Simple to navigate and does not need a separate list of words.

Dale

[igorsk]

One thing that many dictionary formats miss is that one entry can be indexed by many headwords. This is quite important for languages like Japanese. For example, meguirau, 巡り会う, めぐり会う and めぐりあう are all different spellings of the same word and all should match the entry.

[DaleDe]

Originally Posted by igorsk

One thing that many dictionary formats miss is that one entry can be indexed by many headwords. This is quite important for languages like Japanese. For example, meguirau, 巡り会う, めぐり会う and めぐりあう are all different spellings of the same word and all should match the entry.

The entry itself should list all the words that match it IMHO.

Dale