Artifact 59d636749ee66388c329da7e490b3c59f433c948

File todo-ak.txt part of check-in [464a46dad2] - Added a make target to invoke the testsuite, and a make variable for tclsh. Danger. Testsuite modifies the repository the checkout belongs to, and not a copy. Use with caution, clone your development repository and workspace before use. Updated my work list. by aku on 2007-08-26 06:45:02.

[Andreas Kupries]

Things to work on:

* Rework the code underlying the test target to clone the repository
  and workspace before running tests, to avoid modifying the
  development repository by accident.

* Think about exposure of functionality as libraries, i.e. Tcl
  packages. Foundations like delta, etc. first, work up to
  higher-levels.

* Document delta format, delta encoder.

* Document the merge algorithm.

* Document the xfer protocol.

Questions

* In the timeline seen at http://fossil-scm.hwaci.com/fossil/timeline
  the manifest uuids are links to pages providing additional
  information abut the change, differences, parent/child linkage, etc.

  The timeline accessible generated by 'fossil server' and accessible
  at http://localhost:8080/timeline however does not have such links.

  One wonders why ?

General ideas:

* Use TLS/SSL in the xfer module to enhance privacy of
  communications. Integrity too, however that is also handled in the
  fossil protocol itself. 

  Possibility: OpenSSL.
  Dis: Library is large, makes self-containment of fossil difficult.
  Adv: Known library, proven? design/robustness/implementation? -> Marketing

  Possibility: Tcl 8.5+extended libtom* functionality
  Dis: New implementation, less proven
  Adv: More selfcontained.

  Possibility: lib-gnupg
  Dis:
  Adv:

* Talk to Mark Roseman about his Projectforum system and its
    suitability as the wiki component. And in reverse the possibility
    of using Fossil as the version engine underneath ProjectForum.

* Talk to Colin McCormack / Kevin Kenny about usability of Fossil as
  version engine underneath Tcler's Wiki, and also for the wiki
  itself. (Note: Chat message by kbk showed that he is aware of
  fossil).

Notes:

*  Stephen Huntley - tdelta (tclvfs, versioned filesystem, see also
   http://wiki.tcl.tk/12803), compare against delta here.

* fts2 as is is difficult to use in fossil. Indexing currently
  requires the blob data to be fully visible, counter to the extensive
  use of compression (delta,zlib) in repositories.