[PATCH 4 of 8] transaction: avoid file stat ambiguity only for files in blacklist

[FUJIWARA Katsunori]

At Sun, 2 Jul 2017 23:06:04 +0900,
Yuya Nishihara wrote:
> 
> On Fri, 30 Jun 2017 01:53:11 +0900, FUJIWARA Katsunori wrote:
> > # HG changeset patch
> > # User FUJIWARA Katsunori <foozy at lares.dti.ne.jp>
> > # Date 1498754869 -32400
> > #      Fri Jun 30 01:47:49 2017 +0900
> > # Node ID d4d47784511456e6cf77b194822f23586e052312
> > # Parent  9114ae36f40b733912705ef0bacb2961941f18d6
> > transaction: avoid file stat ambiguity only for files in blacklist
> 
> > diff --git a/mercurial/transaction.py b/mercurial/transaction.py
> > --- a/mercurial/transaction.py
> > +++ b/mercurial/transaction.py
> > @@ -44,11 +44,12 @@ def active(func):
> >      return _active
> >  
> >  def _playback(journal, report, opener, vfsmap, entries, backupentries,
> > -              unlink=True):
> > +              unlink=True, noambigstatfiles=None):
> >      for f, o, _ignore in entries:
> >          if o or not unlink:
> > +            checkambig = noambigstatfiles and (f, '') in noambigstatfiles
> >              try:
> > -                fp = opener(f, 'a', checkambig=True)
> > +                fp = opener(f, 'a', checkambig=checkambig)
> >                  fp.truncate(o)
> >                  fp.close()
> >              except IOError:
> > @@ -101,7 +102,8 @@ def _playback(journal, report, opener, v
> >  
> >  class transaction(object):
> >      def __init__(self, report, opener, vfsmap, journalname, undoname=None,
> > -                 after=None, createmode=None, validator=None, releasefn=None):
> > +                 after=None, createmode=None, validator=None, releasefn=None,
> > +                 noambigstatfiles=None):
> >          """Begin a new transaction
> >  
> >          Begins a new transaction that allows rolling back writes in the event of
> > @@ -110,6 +112,10 @@ class transaction(object):
> >          * `after`: called after the transaction has been committed
> >          * `createmode`: the mode of the journal file that will be created
> >          * `releasefn`: called after releasing (with transaction and result)
> > +
> > +        `noambigstatfiles` is a set of (path, vfs-location) tuples,
> > +        which determine whether file stat ambiguity should be avoided
> > +        for corresponded files.
> 
> IIUC, this is a set of files that could get ambiguous, but "noambig" semed to
> say the opposite.

I named this as "(caller requires) no ambiguity of stat for files (in
this set)". But I agree that this name is not understandable :-)

How about "avoidambigfiles" or "avoidstatambigfiles" ? or any ideas ?


> Other than that, the code looks fine. FWIW, I'm getting a feeling that we've
> failed to design a sensible cache-validation API because it's quite hard to
> rely only on mtime/ctime.

I agree, too. But it also is fact that there is no other better (and
portable) information to achieve exact cache validation :-<

-- 
----------------------------------------------------------------------
[FUJIWARA Katsunori]                             foozy at lares.dti.ne.jp