Tailor

[Lele Gaifax]

>>>>> "Andrew" == Andrew Walrond <andrew at walrond.org> writes:


    Andrew> There has never been a ucore directory; don't know what
    Andrew> thats about.

This was caused by a glitch in the hg backend, now fixed: it was
actually u'core', ie the unicode version of the string "core" that
erroneously reached the shell, that interpreted as "ucore".

    Andrew> The resultant repo also looks confused:

Investigating, I seem to understand that hg does not track
directories; nowhere in the the manual there is a mention of them, and
"hg add directory" gives me "directory does not exist". Is this right?
If so, I'm undecided on which is the right way of replaying a patch
that moves a directory, between "hg addremove" and traversing the
whole subtree doing "hg add" as appropriate...

Anyway, giving "-D -v" flags also to the update command (the second
you ran), reveals more details on what's going on; in particular, this
patch:

Changeset Port device-mapper, fcron and exim:

darcs pull --all --patches="Port device-mapper, fcron and exim" .. [Ok]
darcs changes --patches="Port device-mapper, fcron and exim" --xml-output --summ ................... [Ok]
hg add "core" "core/packages" heretix/core does not exist!
heretix/core/packages does not exist!
 [Ok]
hg copy "core/packages/development/device-mapper.rb" "core/packages/device-mapper.rb" heretix/core/packages/device-mapper.rb does not exist!
 [Ok]
hg remove "core/packages/development/device-mapper.rb"  [Ok]
hg copy "core/packages/development/exim.rb" "core/packages/exim.rb" heretix/core/packages/exim.rb does not exist!
 [Ok]
hg remove "core/packages/development/exim.rb"  [Ok]
hg copy "core/packages/development/fcron.rb" "core/packages/fcron.rb" heretix/core/packages/fcron.rb does not exist!
 [Ok]
hg remove "core/packages/development/fcron.rb"  [Ok]
hg commit -u "andrew at walrond.org" -l /tmp/tailorDgYUM1hg -d '1118822900.0 UTC' [Ok]

The pulled patch is seen like this from tailor:

Revision: Port device-mapper, fcron and exim
Date: 2005-06-15 10:08:20
Author: andrew at walrond.org
Modified: core/packages/device-mapper.rb,core/packages/exim.rb,core/packages/fcron.rb
Renamed: core/packages/device-mapper.rb (from core/packages/development/device-mapper.rb),core/packages/exim.rb (from core/packages/development/exim.rb),core/packages/fcron.rb (from core/packages/development/fcron.rb)
Log:

So, at first tailor seems doing the right steps here, but I do not
understand those "does not exist" messages: all mentioned filenames
*are* there... Am I right transposing a rename operation into a "copy
old new; remove old"? When tailor executes the above "copy+remove"
dance it has already applied the darcs patch, that moved those files,
so actually the source is already missing (but hg complains about
target not existing...). 

Mercurial recorded the above patch as

changeset:   5:d02ca3eb31a3b5d03002cf16e6cf8454ca9e8e1c
tag:         tip
manifest:    5:3ae632b4c1204679130b1f24a350f1e8841e7b68
user:        andrew at walrond.org
date:        Wed Jun 15 10:08:20 2005
files:
description:
heretix: changeset Port device-mapper, fcron and exim

which is double wrong, since a) it seems wrong having a patch that
does not touch any file, expecially when b) it should (at least) have
committed the removes that hg seemed to accept.

BTW, shouldn't hg exit with an error status different from zero, when
it encounter wrong situations?

Maybe I just need to enhance my hg knowledge....

Thanks for reporting the problem, and for any hint on this,
bye, lele.
-- 
nickname: Lele Gaifax	| Quando vivrò di quello che ho pensato ieri
real: Emanuele Gaifas	| comincerò ad aver paura di chi mi copia.
email: lele at seldati.it	|		-- Fortunato Depero, 1929.