The work done for issue 22 didn't properly account for the bookinfo section which lead to it being improperly generated.

On rare occasion it is useful to see the horrible wall of text that the xslt generation produces. We add a -v option which basically tells us to stop redirecting stdout.

It would be nice if we could move away from the crazy java build tool that we have and instead move to something like xsltproc and use the makefile we currently have. This would also make it slightly easier to try and generate other targets like say a pdf or something that could be laid out for imposition. But that's fantasy future land.

The document information should be retooled at some point, but it's not a very useful landing page. Instead the work's preface is much better.

Until index generation works correctly for the default output format we should not include it so as not to confuse people as to why there is an empty index.

Currently there is commented out book sections that basically refer to sun docs, the sun community and mailing lists that couldn't care less about our feedback. However, some of us do actually care about that feedback! So we should replace this section with something useful, or just remove it entirely.

Currently the transformer put errors to stdout which get swallowed up by the shell.

Trying to write a dmod is an exercise in finding one that already exists and being rather annoyed. We should make this clearer.

The current default style has an OpenIndiana logo. We should switch this out to be the illumos logo and update the styles as approrpiate.

We are missing the pdl required appendix. This will fix this and clear up legal questions.

Currently the master form of all the images in the gate are encapsulated post script files. It so happens that we have a copy of them translated into gif files. Rather we should make this translation happen every time so we don't have to keep checked in copies in the gate.

The current SolBook transformer exits zero in the face of errors, sigh.

It currently points to illumos/docbooks/issues/new, not the root of the workspace.

Currently the generated index is empty. Further all of the places that use a tag with '' are actually leaking it through. Rather than being consumed for the index, it's actually being dropped into the section.

Figures aren't displaying though links to them do exist. Presumably there is no xslt transformation defined for them. Sigh.

Currently the way that we generate these names is based on ordering information which is subject to change. Rather for better permanently linking we should make these based on the file name.

opensolaris.org is dead to us.

Since mdb tab completion was integrated there are new module interfaces and other information that should be updated to the document to reflect its presence.

Thanks to @jclulow we have an online version of the manual pages. This means that in addition to the current formatting we should make them all links to the illumos online manual pages.

Currenly, the following XML:

<olink targetdoc="refman2" targetptr="write-2" remap="external">
<citerefentry><refentrytitle>write</refentrytitle><manvolnum>2</manvolnum>
</citerefentry></olink>

turns into 'write2' as opposed to 'write(2)'.

Currently there is no good information on using DTrace inside of a zone. This includes everything from what is available inside of a zone to using the #pragma for USDT.

They're dead Jim.

If a document has parts which contain several chapters several things don't work including:

• Generation of the part introduction
• Navlink generation and highlighting the currently active part

Index generation is rather broken, see issue number four. As a side effect all the indexterm sections are being thrown out directly into the generated html in the

tags. This is rather problematic. For now, we're going to add some xslt to the default transform (Indiana) to strip out anything in there. This would break our index, if it worked. But it doesn't, which is why this happens.

There's an extra copy of the docbook morphing jar that's outside of the dist folder which by this Java code's design pretty much guarantees that it won't work. Because of this, we're just going to get rid of it.

Currently most of the section names are named based on an increasing numerical id. Furthermore the index generation relies on these names for its hotlinks. Rather than using a numeric id which can shift, we should use a name based on the section title.

We currently don't have a good way to enforce xml cleanliness. Instead we should use the techniques from issue 30 in a Makefile target. The Makefile is the preferred option as the long term goal is to not need the original hacky Java source.

All over the place in the DTrace guide it mentions Solaris. But there is no Solaris, only illumos. Update this as appropriately.

We should mention the other OSes that this guide supports and the guides applicability to them.

If you use the example tag then the title gets displayed twice. Once in the proper formatting and before that as an extra bold tag.

Add stuff in README to indicate this.

This includes updating licensing information, verifying output formatting, and making the guide realize that Sun is dead.

Currently the html template is not correctly formatting the manual page sections note that the chapter on dtrace(1M) is being rendered as dtrace1M.

The current dtrace(1M) usage is entirely borken. It looks like:

 dtrace3264 aACeFGHlqSvVwZ bbufsz ccmd Dname=def Ipath Lpath ooutput ppid sscript Uname xarg=val Xacst Pproviderpredicate action mprovider: modulepredicate action fprovider: module: funcpredicate action nprovider: module: func: namepredicate action iprobe-idpredicate action 

This includes updating licensing information, verifying output formatting, and making the guide realize that Sun is dead.

This would mostly cover a discussion of using vmregs, how it actually works, and having to expose the ugly truth of the VMCS.

Since the book was last updated, DTrace has been updated with a lot of new things:

• llquantize
• vmregs
• tracemem takes an optional size arguemnt
• toupper/tolower support
• lltostr taking an optional base argument
• -x temporal
• zone improvements (included the zone option)
• print()

This ticket does not include new providers.

Because DYNMCTRCGGD is a lot easier than dtrace.