Things I've found: * There is no longer any creation time exposed through OS.File on Linux. https://tbpl.mozilla.org/?tree=Try&rev=868deba6a819 -- creationDate quietly returns 0. Instead, I have to try something like this: date = info.creationDate || info.s_st_ctime || info.s_st_mtime; and hope nothing breaks. * macBirthDate doesn't seem to work at all. I'm running on 10.8.2 with HFS+: self._log.debug("INFO: " + JSON.stringify(info)); self._log.debug("CREATION: " + info.creationDate); self._log.debug("BIRTH: " + info.macBirthDate); 0:00.40 INFO: {"isDir":false,"isSymLink":false,"size":0,"creationDate":"2013-01-06T19:08:14.000Z","lastAccessDate":"2013-01-06T19:08:14.000Z","lastModificationDate":"2013-01-06T19:08:14.000Z","unixLastStatusChangeDate":"2013-01-06T19:08:14.000Z","unixOwner":501,"unixGroup":501,"unixMode":420} 0:00.40 CREATION: Sun Jan 06 2013 11:08:14 GMT-0800 (PST) 0:00.40 BIRTH: undefined This doesn't make any sense at all: // Deprecated, use macBirthDate/winBirthDate instead get creationDate() { // On the Macintosh, returns the birth date if available. // On other Unix, as the birth date is not available, // returns the epoch. return this.macBirthDate || new Date(0); }, implies that if I'm getting `undefined` there, I should be getting epoch for creationDate. What's going on?

Further: self._log.debug("macBirthDate present? " + ("macBirthDate" in info)); > 0:00.58 macBirthDate present? false

> * There is no longer any creation time exposed through OS.File on Linux. That's by design. The information is simply not available at system level (see e.g. http://linux.die.net/man/2/fstat) So WONTFIX for this part of the bug. > * macBirthDate doesn't seem to work at all. I'm running on 10.8.2 with HFS+: [...] Mhhh... Sounds like there is a problem with _DARWIN_FEATURE_64_BIT_INODE. I'll take a look. How urgent is that for you?

(In reply to David Rajchenbach Teller [:Yoric] <away until Jan 7th> from comment #2) > > * There is no longer any creation time exposed through OS.File on Linux. > > That's by design. The information is simply not available at system level > (see e.g. http://linux.die.net/man/2/fstat) > > So WONTFIX for this part of the bug. So this is the best course of action? function onStatSuccess(info) { // OS.File doesn't seem to be behaving. See Bug 827148. // Let's do the best we can. This whole function is defensive. let date; if ("winBirthDate" in info) { date = info.winBirthDate; } else if ("macBirthDate" in info) { date = info.macBirthDate; } if (!date || !date.getTime()) { self._log.debug("No birth date: using ctime/mtime."); try { date = info.creationDate || info.lastModificationDate || info.unixLastStatusChangeDate; } catch (ex) { self._log.debug("Exception fetching creation date: " + ex); } } if (date) { let timestamp = date.getTime(); self._log.debug("Using date: " + entry.path + " = " + date); if (timestamp < oldest) { oldest = timestamp; } } > Mhhh... Sounds like there is a problem with _DARWIN_FEATURE_64_BIT_INODE. > I'll take a look. How urgent is that for you? If it affects the accuracy of the returned value, we will need a fix landed in Aurora for FHR.

(In reply to Richard Newman [:rnewman] from comment #3) > (In reply to David Rajchenbach Teller [:Yoric] <away until Jan 7th> from > comment #2) > > > * There is no longer any creation time exposed through OS.File on Linux. > > > > That's by design. The information is simply not available at system level > > (see e.g. http://linux.die.net/man/2/fstat) > > > > So WONTFIX for this part of the bug. > > So this is the best course of action? Something along the lines of: date = info.winBirthDate || info.macBirthDate || info.lastModificationDate > > Mhhh... Sounds like there is a problem with _DARWIN_FEATURE_64_BIT_INODE. > > I'll take a look. How urgent is that for you? > > If it affects the accuracy of the returned value, we will need a fix landed > in Aurora for FHR. I will work on it this week.

Ok, this should be fixed by bug 802534.

Richard, can you confirm that the new behavior works for you?

(In reply to David Rajchenbach Teller [:Yoric] from comment #6) > Richard, can you confirm that the new behavior works for you? I don't see a concise summary of what the new behavior is, beyond Comment 4, so my only answer is "I think so"! We do this: function onEntry(entry) { function onStatSuccess(info) { let date = info.winBirthDate || info.macBirthDate; if (!date || !date.getTime()) { self._log.debug("No birth date. Using mtime."); date = info.lastModificationDate; } if (date) { let timestamp = date.getTime(); self._log.debug("Using date: " + entry.path + " = " + date); if (timestamp < oldest) { oldest = timestamp; } } } and our basic tests pass.

Closing the test, then.