From Fedora Project Wiki

No edit summary
Line 57: Line 57:
Publican produces the rpm very differently than the way we used to.  For F11, we more or less followed the model of F10 with the single change that the release notes (only one of 6 docs in release-notes.rpm) were built in the srpm.  Previously, already built docs were provided.
Publican produces the rpm very differently than the way we used to.  For F11, we more or less followed the model of F10 with the single change that the release notes (only one of 6 docs in release-notes.rpm) were built in the srpm.  Previously, already built docs were provided.


Currently, the release notes and the supporting readmes are supplied in the RPM in all the languages (currently around 40).  These are installed on every system.  The Publican model is to have one document in one language per RPM.  It had been suggested that we could subpackage the languages, allowing only the installation of the relevant language.  While this could save considerable space, especially as the number of languages increases, it is not at all clear that current release engineering practices and tools could support this in actual practice.
Currently, the release notes and the supporting readmes are supplied in the RPM in all the languages (currently around 40).  These are installed on every system.  The Publican model is to have one document in one language per RPM.  It had been suggested that we could subpackage the languages, allowing only the installation of the relevant language.  While this could save considerable space, especially as the number of languages increases, it is not at all clear that current release engineering methods and tools could support this in actual practice.
 
=== Can we get a cheat sheet ===
 
Currently, tagging in the wiki is very uneven, and not always translatable into xml markup without a lot of knowledge on the part of the author.  We have all sorts of pages on how to write for the wiki, but these tend to be long, rambling prose that may be reasonable to read, but easy to forget.
 
What we need is a single page that lists wiki markup and corresponding xml markup, with no excess flowery prose to confuse the issue.  This could become a quick reference for authors that doesn't take the energy that the current documents require to pull useful nuggets.
 
Along with this we might see if we can move toward markup that is more easily reflected in xml (at least for release notes where the ultimate target is xml).  Some markup (e.g. <code>) translates directly to xml, while others (templates in particular) ate very error-prone when making the conversion.
 
 


[[Category:Docs_Project]]
[[Category:Docs_Project]]

Revision as of 14:50, 1 May 2009

There are a number of decisions about how the Docs team will deal with F12 that we should be thinking about.

I have much to add to all of these topics, please feel free to expand on any of these, and add additional decisions we should make.

What will release notes look like

There has been some discussion that the release notes are too long. On the one hand, more detailed release notes discourage people from reading them and are more work for translators. On the other, every change is important to somebody.

A proposal I would like to make is as follows:

  • Continue to have the beats follow the release notes pretty much as it is now
  • For each beat, emphasize one or two really cool new features
  • Following that, list each changed package with the old and new versions and a link to the upstream page

That way someone waiting for a particular bugfix can easily determine whether it is there, the release notes are shorter and much more approachable, and a lot less material needs to be translated.

What will we use for language-loc codes

Fedora, ISO, and publican all have different ideas about language-location codes. What will we do?

  • Fedora uses <lang>_<loc> with _<loc> missing when it is not available.
  • Publican uses <lang>-<loc> always
  • ISO indicates one should use <lang>-<loc> with -<loc> missing if it is not relevant

In Fedora 11, the release notes were produced using the Publican convention. However, release notes are displayed with yelp, which uses an .omf file to determine which language to display. .omf files are provided for both the Fedora and Publican conventions so that the release notes always display in the correct language.

The Docs Project could decide to simply produce documents following the Fedora standard. This would be simple to do and not involve any other teams. Publican appears to be happy with whatever languages it is handed in the Makefile.

It might make more sense to attempt to move the entire Fedora project to the ISO standard. This would involve many other teams. Changes to Publican would be nice, but not required.

How should release notes be displayed

Currently, release notes and about fedora are displayed using yelp. The remaining documents are left for the user to discover in /usr/share/doc/HTML. What is the right answer?

Currently, there are menu choices for "About Fedora" and "Help". "About Fedoa" opens the about-fedora doc in yelp. "Help" provides a selection which includes about-fedora and fedora-release-notes (along with some Gnome information which seems to change from release to release).

There appear to be no menu choices leading to README, README-burning-isos, README-accessibility and README-live-images.

By default, the Publican produced RPM adds a "Documentation" menu above "Help" and places the document on that menu. The expectation is that the document is in HTML and is viewed with a browser or HTML viewer. Yelp documents are in XML.

Should we convert minor docs to Publican

README, README-burning-isos, README-live-images, homepage and About-Fedora are still produced using fedora-doc-utils. Should they be converted to Publican?

What is the future of homepage

It isn't entirely clear that this is even needed. Apparently, once upon a time in the distant past, an offline user was directed to homepage rather than seeing a network failure error. This is apparently no longer the case (although there doesn't seem to be total agreement on whether it is or is not). There appears to be no other way to end up at this page other than navigating /var/doc/HTML/...

This page may simply be excess baggage, but some may view the Firefox behavior of failing to end up at this page when the user is offline as a bug.

What does release-notes.srpm look like

Publican produces the rpm very differently than the way we used to. For F11, we more or less followed the model of F10 with the single change that the release notes (only one of 6 docs in release-notes.rpm) were built in the srpm. Previously, already built docs were provided.

Currently, the release notes and the supporting readmes are supplied in the RPM in all the languages (currently around 40). These are installed on every system. The Publican model is to have one document in one language per RPM. It had been suggested that we could subpackage the languages, allowing only the installation of the relevant language. While this could save considerable space, especially as the number of languages increases, it is not at all clear that current release engineering methods and tools could support this in actual practice.

Can we get a cheat sheet

Currently, tagging in the wiki is very uneven, and not always translatable into xml markup without a lot of knowledge on the part of the author. We have all sorts of pages on how to write for the wiki, but these tend to be long, rambling prose that may be reasonable to read, but easy to forget.

What we need is a single page that lists wiki markup and corresponding xml markup, with no excess flowery prose to confuse the issue. This could become a quick reference for authors that doesn't take the energy that the current documents require to pull useful nuggets.

Along with this we might see if we can move toward markup that is more easily reflected in xml (at least for release notes where the ultimate target is xml). Some markup (e.g. <code>) translates directly to xml, while others (templates in particular) ate very error-prone when making the conversion.