doc/book/TODO

This is a -*- text -*- file.

Here is a list of all the things that the Subversion Book DOESN'T yet cover.

--------------------------------------------------------------

Stuff that needs more discussion:

  - 'svn diff/merge --ignore-ancestry'.  Demonstrate how this switch
    can be used as an alternate strategy for vendor branching (instead
    of using svn_load_dirs.pl): just import each new vendor release
    unrelatedly, and apply the diffs to your vendor-branch working copy.

  - (perhaps) suggest the use of environment variables to represent
    long repository URLs.  useful for lots of branching/tagging commands.

--------------------------------------------------------------
To be fixed by someone:

  - svnadmin docs sorely need to be updated.  See Ch05 and Ch08.

--------------------------------------------------------------

To be fixed by cmpilato:

  - define a 'project root', recommend 'standard' repos layout beneath
    such roots.

  - svndumpfilter.  in particular, show how to "migrate" a subdir in
    one repos to a subdir in another one.  Wherever this ends up
    documented (probably in chapter 5 or 6), please be sure to mention
    how to use svndumpfilter in chapter 4, "removing a change from the
    repository".  At a minimum, create a xref from this part of
    chapter 4 to the *real* svndumpfilter documentation, presumably to
    use-case that shows how to use svndumpfilter as an alternative to
    'svnadmin obliterate.'

  - ch07: maybe a explanation for the person writing a new main()
    application... what are all those callbacks in the ctx_t used for?
    people have asked this in irc.  what sort of things does an
    application *have* to set up in the cxt_t in order to call an
    svn_client_foo() function?

  - I think we need to add  a *bit* more background on "how berkeley db"
    works.  It's not clear to most readers 1) what the db logfiles are,
    2) that they need to be pruned regularly, 3) that they grow *even*
    if you only do checkouts and updates.   Ben wrote an email about
    this which should probably be assimilated into ch05:

      http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=38769

    While it's not our job to turn the readers into BDB experts, I
    think this level of information is the exact right level of BDB
    knowledge that users/admins need to use svn well.

--------------------------------------------------------------

To be fixed by fitz:

  - appendix A: how authentication differs from CVS.  Explain that the
    server challenges, rather than the client explicitly "logging in".

    Also, in 'branches are dirs' section, stress that users would NOT
    want to ever checkout the root of an svn repository.... it would
    be humongous.  Some CVS users have been confused by this.

  - some folks want a glossary.  that's good, since we have a
    glossary.xml anyway.  :-)  probably something we do at the last minute.

  - Check all commands for flags in ch08

--------------------------------------------------------------

To be fixed by sussman:

  - certificate support in the client.  how to configure
    ~/.subversion/servers, manage certs, etc.

  - the auth data now lives in ~/.subversion/auth.  explain exactly
    when stuff gets cached, how to prevent it from being cached, and
    how to invalidate a cache.

  - new 'svn diff' UI (when it's done) -- affects ch03, ch08, and
    maybe other chapters.

  - svn merge --dry-run   (different than 'svn diff')

  - ch06: document the [tunnels] section of ~/.subversion/config.
    also document the log-encoding and use-commit-times variables in
    the [miscellany] section

  - document mod_authz_svn!  and don't forget to mention it in the
    'svnserve vs. apache' comparison section of ch05.

--------------------------------------------------------------

To be fixed by kfogel:

  - cvs2svn.py is a lot more complex now, with many more switches,
    especially since it now converts branches and tags.  Need to
    expand our coverage of this in Appendix B.