sfeed_tail: improvements

- sfeed_tail only reads from files now, not from stdin anymore. This had too many
  caveats.
- Instead of a timer it now detects changes faster and is more efficient using
  memory.
- Improve documentation of its behaviour.

1 files changed, 14 insertions, 21 deletions

diff --git a/sfeed_tail.1 b/sfeed_tail.1
index a64aca1..9734339 100644
--- a/sfeed_tail.1
+++ b/sfeed_tail.1
@@ -1,4 +1,4 @@
-.Dd January 29, 2016
+.Dd August 22, 2018
 .Dt SFEED_TAIL 1
 .Os
 .Sh NAME
@@ -6,40 +6,33 @@
 .Nd format unseen feed data to a plain-text list
 .Sh SYNOPSIS
 .Nm
-.Op Ar file...
+.Ar file...
 .Sh DESCRIPTION
 .Nm
-formats unseen feed data (TSV) from
-.Xr sfeed 1
-from stdin or
+formats only new and unseen feed data (TSV) from
+one or more
 .Ar file
 to stdout as a plain-text list.
-If one or more
-.Ar file
-are specified, the basename of the
+The basename of the
 .Ar file
 is used as the feed name in the output.
-If no
-.Ar file
-parameters are specified and so the data is read from stdin the feed name
-is empty.
-.Pp
-.Nm
-will mark the initial items on the first run as seen.
-Then it will print the initial items if data is read from stdin, if the data
-is read on the first run by specifying the argument(s)
-.Ar file
-it will not show the initial seen items.
 .Pp
 Unseen items are printed per line in a similar format to
 .Xr sfeed_plain 1 ,
 duplicate items are ignored.
-The list of unique items is determined by the fields: feedname, item id and
+The list of unique items is determined by the fields: item id, item link and
 UNIX timestamp of the item date.
+.Pp
+.Nm
+will also only process and show items that are considered new: the item
+timestamp is not older than a day ago.
 .Sh IMPLEMENTATION NOTES
+.Nm
+checks for file modifications each second by checking the filesize and
+modification time.
 Keep in mind that because
 .Nm
-keeps a list of all the items it can consume much memory.
+keeps a list of items it can potentially consume much memory.
 .Sh SEE ALSO
 .Xr sfeed 1 ,
 .Xr sfeed_plain 1 ,