Repository navigation
borg2: list: Display nanosecond precision timestamps #10583
Description
https://github.com/borgbackup/borg/commit/6f9cb17c80043fb3a32bc8cc77642f9969e1edea provided the possibility to display timestamps in nanosecond precision. However, the feature doesn't seem to be used, e.g. in borg list. I propose a small patch to actually display nanosecond timestamps:
--- /tmp/a/parseformat.py 2026-11-01 17:00:22.573067730 +0100
+++ /tmp/b/parseformat.py 2026-11-01 17:04:15.289401967 +0100
@@ -29,7 +29,7 @@
from .fs import make_path_safe, slashify
from .argparsing import Action, ArgumentError, ArgumentTypeError, register_type
from .msgpack import Timestamp
-from .time import OutputTimestamp, format_time, format_time_ns, safe_timestamp
+from .time import OutputTimestamp, format_time, format_time_ns, safe_timestamp, safe_ns
from .. import __version__ as borg_version
from .. import __version_tuple__ as borg_version_tuple
from ..constants import * # NOQA
@@ -1355,7 +1355,8 @@
return hash.hexdigest()
def format_time(self, key, item):
- return OutputTimestamp(safe_timestamp(item.get(key) or item.mtime))
+ ts = item.get(key) or item.mtime
+ return OutputTimestamp(safe_timestamp(ts), safe_ns(ts)%1_000_000_000)
def format_iso_time(self, key, item):
return self.format_time(key, item).isoformat()
Reacted by TW
Activity
ThomasWaldmann commentedon Nov 1, 2026
Well, this has pros and cons:
I am somehow indifferent, could live with both ways.
So guess this needs more feedback from other users:
PhrozenByte commentedon Nov 1, 2026
Regarding
borg listspecifically: We could populate the timestamps with nanosecond precision but use{mtime:24}in the default--formatstring (i.e., hide them by default).Regarding nanosecond precision in general: IMHO, normal text output is intended for humans, and humans rarely (if ever) care about nanoseconds. Nanoseconds should be included in
--jsonoutput, though. Also, see my earlier thoughts on this in #9840 (comment)goebbe commentedon Nov 2, 2026
Personally, I never missed nanoseconds in
borg listfor my own usage, with a relative small number of backups.What would be the real world use case, where nanoseconds in
borg listmatter?Would this type of precission be useful for real world debugging/ professional usecases?
To me it seems that log files are a better place for debugging/ precise time stamps.
Anyway, if nanoseconds are added, one should think about good default formating. On second is 1.000.000.000 nanoseconds - and I guess nobody wants to spend time e.g. searching for minutes and seconds in long time stamps, in the output of
borg list.mburghardt commentedon Nov 2, 2026
My personal use case is to compare the output of
borg list--json...against a database that stores the file properties / checksums at the time of archive creation in order to ensure data integrity. As the originating filesystem provides nanosecond precision it seems only natural to me to use it, especially as Borg has the data internally.However, I agree with @PhrozenByte and @goebbe to differentiate between output like JSON or log files and text output that is intended to be read by humans.