[Mldonkey-users] [patch #4649] Environment variables for file_completed_cmd

[spiralvoice]

Follow-up Comment #6, patch #4649 (project mldonkey):

"- should environment variables be consistent across all spawned processes?
(see: file_start_command)"
there is no problem extending file_start_command, suggestions welcome.

"I assume it currently could move directly to the quarantine from the
pre-commit command."
I don´t know if MLDonkey handles this case properly, this means that a
finished file is removed from temp before commit. file_commit does lots of
important things which could break if the file is removed, but I never tested
this case. I am even thinking about removing auto_commit option and commit all
finished files immediatly, what about that? There are other problems with
finished, but not committed files - BT files don´t upload until they are
committed. Now there are two choices, fix this bug in BT or commit all files
immediatly. This change should be subject to discussion to know how many
users really de-activate auto_commit and why. Pango suggested that users,
where incoming and temp are on different partitions, would like to control
the diskspace needed for committing files. As MLDonkey will control this
itself in the future this would make using
auto_commit false obsolete in this case.

"Is there a problem with having both pre and post commit command options?"
No, unless the filename for the committed file is omitted there because
pre-commit command can never provide a 100% correct filename due to name
collisions with other files in incoming when commit takes place.

    _______________________________________________________

Reply to this item at:

  <http://savannah.nongnu.org/patch/?func=detailitem&item_id=4649>

_______________________________________________
  Nachricht geschickt von/durch Savannah
  http://savannah.nongnu.org/