Some feeds never have any content in their feed, so they would have a SHA1 reference of da39a3ee5e6b4b0d3255bfef95601890afd80709 for all their entries.
elfeed-deref would look up this hash in the compressed archive first, will likely find it, decompress the archive only to return an empty string (car index) == (cdr index). This is quite some overhead only to return an empty string. I'd propose to make elfeed-ref return early whenever we face a hash that represents an empty string.
For what it's worth, I worked around it by advising elfeed-deref with the following code:
(defun bram85-elfeed-deref-optimization (f &rest args)
"Optimize for empty entry content.
This is an advice meant for `elfeed-deref' (argument F) with a
reference ID in ARGS. This advice intercepts a reference to empty
content and immediately returns the empty string. `elfeed-deref'
would likely find the empty content inside archive.gz, decompress
it and only then return an empty string.
Some feeds structurally have empty content, inducing a lot of
overhead otherwise."
(let ((sha1-empty-string "da39a3ee5e6b4b0d3255bfef95601890afd80709")
(ref-id (elfeed-ref-id (car args))))
(if (equal ref-id sha1-empty-string)
""
(apply f args))))
(advice-add 'elfeed-deref :around #'bram85-elfeed-deref-optimization)
Some feeds never have any content in their feed, so they would have a SHA1 reference of da39a3ee5e6b4b0d3255bfef95601890afd80709 for all their entries.
elfeed-derefwould look up this hash in the compressed archive first, will likely find it, decompress the archive only to return an empty string(car index) == (cdr index). This is quite some overhead only to return an empty string. I'd propose to make elfeed-ref return early whenever we face a hash that represents an empty string.For what it's worth, I worked around it by advising
elfeed-derefwith the following code:(Source)