Comment by anthk | Hacker News Reader

Comment by anthk | Hacker News Reader

Parent Full thread

Gnus it's dog slow, be with email or with usenet. I say this as an ex-Emacs user where I even plugged slrn's cache in order to speed up things, but over 100k messages that was unusable in my netbook, even under 64 bit machines and native compilation. Slrn did it better. On RSS, I use sfeed which is more Unix like and I just plumb lynx/links or whatever I like as a reader. And fast, much fast than GNUs, Elfeed or the core RSS reader in Emacs.

OTOH, Emacs it's the only libre Usenet reader for Android. Go figure, and that being a dead simple protocol. Despite of that, lots of Thunderbird forks in FDroid didn't adapt the Usenet part yet.

Offpunk it's slow but adding multiprocessing with flock (for python3 maybe) would be a piece of cake in order to allow parallel downloads while syncinc.

offpunk’s slowness has two main sources (while offline):

- loading modules at startup (will be solved in 3.1 with lazy loading, patch is pending)

- parsing HTML with lot of pictures (because we wait for chafa for each picture)

I’m not sure how multiprocessing could really help that much but I would be interested.

While online, sure, the blocking http calls are something that will be parallelized in the future

syncing URL lists should be done in parallel. Something like xargs+flock but under Python.

Contributions are welcome. I even believe that this whole code should be migrated from python-requests to curl.

I don't think Gnus is that bad once you spend some time setting it up. For groups with a ton of content where I mostly want to search, I found it was better to just download the whole group and index into notmuch. I could query 20 years of the Smalltalk USENET group or the Supercollider mailing list instantly.

Read again, I used slrnpull's cache. Mail was fast with Mu4e and Mairix. GNUS with slrn, not much. I have tons of articles from comp.* and some alt.* related groups. People loves to talk at comp.misc and comp.arch.

Also GNUS' caching it's really slow compared to slrn. Yes, I know, Elisp vs C, but even with native compilation it was excruciatingly slow. I've seen bytecoded TCL back in the day running really fast for these tasks.

Reply on news.ycombinator.com