leo/doc/leoNotes.txt now contains prototypes of several bzr hooks. See:
leoPy.leo#Notes-->@file ../doc/leoNotes.txt-->bzr hooks: prototypes
I'll show them in the Post Script.
This is the second time I've played around with bzr hooks. It's not a fun
experience. Both the documentation and the actual bzr hook architecture
are poor. The only way to get something that works is with trial and error
and liberal use of dir to find out what bzr classes actually contain!
This process revealed two big problems:
1. Bzr has no pre_push hook! This would be the hook that calls just before
doing a push. Having a post_push hook seems useless.
2. As I was thinking about this problem, I realized (Doh!) that bzr doesn't
push actual files--it pushes diffs, or their equivalent.
It's a good thing that I didn't know about these problems initially: I
would never have had the original Aha ;-)
I did have the feeling, though, that things might not be all that easy.
Several possible ways forward may be possible:
1. A post_commit hook. This would do the actual "switcheroo" of the
files. But that creates it's own problems. Nancy will see the sentinels
until (presumably) the post_push hook undid the switcheroo. And
switcheroos could well confuse bzr.
2. A completely different alternative would be to create, say, Emacs
load/save hooks. The load hook would strip Leo sentinels, the save hook
would put them back, using the @shadow algorithm. This might be, in some
sense, a bit simpler because bzr is not involved. The implementation would
require that the leoBridge be installed (not a huge deal). And the actual
push/pull hooks would be written in elisp. Again, not a huge deal: all the
real work would be done in the bridge.
Of course Point 2 requires load/save hooks for vim, Eclipse, etc. I
suspect the work involved would be similar to supporting all kinds of
SCCS. Anyway, we only need to do *one* set of load/save hooks as a
prototype.
Edward
P.S. Here are the three prototype files. The traces reveal how little the
docs help. I did most of my exploration using the status hook, because bzr
status can be invoked without affecting the repo.
BTW, it would be reasonable to put all the following code into a single
file, say leo_push_pull_hooks.py.
===== status_hook.py:
from bzrlib import status
def post_status_hook(args):
print('status_hook.py: post_status_hook')
if 0:
# args.old_tree, args.new_tree, args.to_file,
# args.versioned, args.show_ids, args.short, args.verbose
print('\n'.join(
['%s %s' % (z,getattr(args,z)) for z in sorted(dir(args))
if not z.startswith('_')]))
if 0: # disable
status.StatusHooks().install_named_hook(
'post_status',post_status_hook,'EKR post_status hook')
===== push_hook.py:
from bzrlib import branch
def post_push_hook(arg):
print('post_push_hook')
# print('\n'.join(dir(arg)))
if 0: # Disable.
branch.Branch.hooks.install_named_hook(
'post_push', post_push_hook,'EKR post_push hook')
===== command_hook.py
from bzrlib import commands
from bzrlib import mutabletree
def pre_command_hook(*args,**keys):
print('command_hook.py: pre_command_hook')
def post_commit_hook(args):
print('command_hook.py: post_commit_hook')
if 0:
print('\n'.join(
['%s %s' % (z,getattr(args,z)) for z in sorted(dir(args))
if not z.startswith('_')]))
if 0: # does not work.
commands.CommandHooks().install_named_hook(
'pre_command',pre_command_hook,'EKR pre_command hook')
if 0: # works.
mutabletree.MutableTreeHooks().install_named_hook(
'post_commit',post_commit_hook,'EKR post_commit hook')
All in all, a pretty miserable experience.
EKR
--
You received this message because you are subscribed to the Google Groups
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/leo-editor.
For more options, visit https://groups.google.com/groups/opt_out.
