Hi,

here’s a small summary of the 11th (coding) week. This week I spent
most of my time with working on the wc code.

DONE:

• project wc: added commit and update methods
• lots of wc code refactoring

TODO:

• project wc: commit only specific files for a package instead of the
  complete package (the package wc class already supports this)
  (use case: osc ci pkg1/file pkg1/foo pkg2/bar pkg3)
• convert old working copies to the new format
• package wc: update: add support to specify stuff like “expand”, “linkrev”
  etc.
• project wc: add a revert method (to restore a package wc with state ‘!’)
• project/package wc: support diff
• package wc: implement a pull method (does the same as “osc pull”)

Marcus

Hi,

here’s a small summary of the 10th (coding) week. This week I spent
most of my time with working on the package working copy class’
commit method. This also involved some code refactoring.
Basically commit and update are handled as “transactions” and all
relevant transaction data is stored in a single xml file
(pkg/.osc/_transaction/state). This way an update/commit is more or
less an atomic operation.
For more information have a look at [1] (class XMLTransactionState
+ subclasses and commit/update method).

TODO:

• add update and commit methods for the project class

Marcus

[1] https://gitorious.org/osc2/osc2/blobs/master/osc/wc/package.py

Hi,

here’s a small summary of the 8th (coding) week. This week I spent
most of my time with rewriting the working copy code.
Done:

• added support to add and delete packages
• added some “abstractions” for the tracking file format:
  currently packages and files are tracked in a xml file
• thought about the package update algorithm. Basically
  it’ll work like this (verfy simplified version):
  – perform update in a tmpdir (phase 1)
  – if the tmp update finished, copy/rename all files to
  the wc (phase 2)
  Advantage:
  If the update is interrupted in phase 1 the wc wasn’t touched
  at all and nothing should be broken.
  If the update is interrupted in phase 2 the wc is _inconsistent_
  but a subsequent “update” call can resume the update and everything
  should be consistent afterwards (in this case only files are
  copied/renamed)

TODO:

• implement update + commit algorithm

If everything works as expected most parts of working copy code
cleanup should be finished after this week.

Marcus

Hi,

here’s a small summary of the 7th (coding) week. This week I spent
most my time with the project and package classes which manage osc’s
working copies.

Done:

• basic working copy layout
• checks to detect broken/corrupt working copies
• locking support (in order to lock a working copy (for instance when
  doing a commit or an update))

TODO:

• add “core” methods like update, commit, diff etc.
• (auto-) repair methods (to fix broken working copies)

Marcus

Hi,

here’s a small summary of the 6th (coding) week. Unfortunately I had
to spent more time with university stuff than I expected – that’s why
I didn’t finish the complete todo for this week.
I did some code restructuring and started to work on the class for the
source route.
TODO:
– rewrite the project and package working copy classes:

• the new working copy format will be incompatible with the current
  format.
• the basic layout will look like this:
  prj
  |
  —> .osc/ (stores prj _and_ pkg metadata)
  |
  —> pkg1
  #       |
  #        —> <files>
  …
  |
  —> pkgN
  #       |
  #        —> <files>
  So all metadata is stored in the prj/.osc dir instead of prj/pkg/.osc
  The advantage is that we can support a complete package
  “restore” (without the need to download the package again):
  cd prj; rm -r pkg; osc revert/restore pkg;
  (that’s possible because the metadata is stored in the prj/.osc
  dir).
• to convert old project/package working copies to the new format
  the “osc repairwc” command can be used (at least that’s the plan)

Feedback is always welcome.

Marcus