#dev 2021-10-04
2021-10-04 UTC
[timothy_chambe], SamWilson[m], samwilson, Guest2366, maxwelljoslyn[d], [fluffy], alex11 and hendursaga joined the channel
# 
capjamesg[d] I thought I had figured out recrawling for IndieWeb Search but I haven't quite done so yet...
# capjamesg[d] Right now, I collect all feeds on a site and save them to a feeds.json file.
# capjamesg[d] Example:
{"jamesg.blog": [{"url": "https://jamesg.blog/feed.xml", "mime_type": "application/atom+xml", "etag": "NOETAG", "discovered": "2021-10-01"}]}# capjamesg[d] I have two issues. First, updating. Every time I want to update this JSON file, I need to read it, amend the site's feeds list, and then save the file again. Second, this code is multithreaded. What happens if two threads need to read / write at the same time?
# capjamesg[d] I would love to do this without a database because that is more overhead to set up.
# capjamesg[d] I suppose I could technically have a microservice whose sole job is just to save / retrieve feeds that is backed by a database?
# capjamesg[d] Any better alternatives would be greatly appreciated! I have been scratching my head about this for days.
[KevinMarks] and hendursa1 joined the channel
# 
nekr0z capjamesg[d] you need blocking on the file write (or perhaps read/write, depending on your app's architecture) if the app is mutli-threaded, I doubt there's another way. A separate microservice is a valid idea, but it basically achieves the same thing. Would have been the same with a database, though ;)
tetov-irc joined the channel
# capjamesg[d] Thanks for your help everyone!
# capjamesg[d] I am going to just use a database, probably MySQL.
# capjamesg[d] It will be a lot easier and scalable than looking through a JSON file that could get massive 😄
# capjamesg[d] I had no idea you could keep a lock on what process owns a file. That could be super useful sknebel++ / nekr0z++
# capjamesg[d] sknebel++
Seirdy, akevinhuang, [Grant_Hutchins], Ramon[d] and joshproehl joined the channel
hendursa1 joined the channel
# 
[tantek] Is it worth defining “skeleton screen” as an anti-pattern to avoid? https://twitter.com/scottjehl/status/1445024832035463171
# 
@scottjehl Skeleton screens were an interesting idea, but I feel like they've shown to be less a design pattern, more an anti-pattern. I wonder how often they resolve to layouts they resemble. Are these old site layouts? Probably rather see a spinner than a random gray-box page layout step https://pbs.twimg.com/media/FA2-DDeXEAsqP6b.png (twitter.com/_/status/1445024832035463171)
[chrisbergr] joined the channel
# 
jeremycherfas I have a problem loading a module in Python. I know the module is at `~/opt/anaconda3/pkgs/` My project is at `~/notebooks/test/` I tried `.opt.anaconda3.pkgs import geojsonio` and `..opt.anaconda3.pkgs import geojsonio` and got two different errors. is there a Python euqivalent of `~`? Should I be spelling out `user`?
# capjamesg[d] I haven't seen any IndieWeb examples. I know a few big sites do it though. I'm not sure if it's worth defining it as an antipattern for personal sites.
# jeremycherfas I've been reading stuff about importing from packages and modules with a path, and those seemed to be the suggestions.
# capjamesg[d] What errors did you get?
# jeremycherfas But I am probably misunderrstanding.
# capjamesg[d] Can you remove the . and say from opt.anaconda3.pkgs import geojsonio
# capjamesg[d] +1
# jeremycherfas No module named '__main__.opt'; '__main__' is not a package
# jeremycherfas Without the . `ModuleNotFoundError: No module named 'opt'`
# capjamesg[d] sknebel Very true.
# capjamesg[d] Disregard my suggestion.
# jeremycherfas I hope so. I am using Jupyter from Anaconda. And I installed geojsonio using `conda install`
# jeremycherfas No. `ModuleNotFoundError: No module named 'geojsonio'`
# jeremycherfas But if I search for geojsonio it is in `opt/anaconda/pkgs`
# jeremycherfas `from sys.path import geojsonio` gives `ModuleNotFoundError: No module named 'sys.path'; 'sys' is not a package`
hendursaga joined the channel
# jeremycherfas Ahah. So sys.path is showing that all the paths are to the specific anaconda environment that I set up. Let me check site-packages for that.
chenghiz_ joined the channel
# jeremycherfas geojsonio is not there! Thank sknebel++
# jeremycherfas Now, I wonder how I install it into that location, given that Anaconda Navigator does not find geojsonio as a package it can install.
# jeremycherfas goes off to find out how to tell conda where to put things.
# jeremycherfas I'll try that from the Notebook. There is also this https://stackoverflow.com/questions/57326043/how-to-install-packages-in-conda-that-are-not-available-in-anaconda
stephenajulu and kogepan joined the channel
# capjamesg[d] !tell [tantek] I have made a public query that shows a breakdown of feeds found in crawls: https://indieweb-search.jamesg.blog/stats
# capjamesg[d] This is more of a heads up than anything else because the data is not backdated. It will only show new feeds encountered by the index / those found in recrawls.
# capjamesg[d] In a few weeks, the numbers should be much higher 🙂
# capjamesg[d] feed / websub are reserved names, not the actual MIME type found.
mackeveli_, diegov, vikanezrimaya and unrelentingtech1 joined the channel
# capjamesg[d] Why thank you!
mambang[m] joined the channel
# capjamesg[d] Thanks. I have finally thought of a way to make reindexing happen without me. I just need to make sure it works in the field.
unrelentingtech1, nekr0z and plindner[m] joined the channel
# 
Loqi GWG: [tantek] left you a message 1 day, 11 hours ago: I rewrote the h-feed discovery section a bit, hopefully to make it clearer and provide more explicit steps (rather than a long sentence). LMK what you think: https://microformats.org/wiki/h-feed#Discovery
npd[m] and cadeyrn[d] joined the channel
# [tantek] aaronpk, do your notes not display mentions? pretty sure I sent you one 🙂 https://aaronparecki.com/2021/10/04/9/my-blog
# Loqi [Aaron Parecki] you know what's *not* down right now?
my blog ⬇
https://aaronparecki.com/
# [tantek] I mean even twitter's quote tweets feature shows it: https://twitter.com/aaronpk/status/1445072111098220551/retweets/with_comments
# 
@aaronpk you know what's *not* down right now? my blog ⬇ https://aaronparecki.com/ (twitter.com/_/status/1445072111098220551)
# [tantek] aside: for those of you interested in the open/public opportunities to participate in W3C's annual TPAC, registration is now open! https://www.w3.org/2021/10/TPAC/
kogepan joined the channel
# [tantek] aaronpk, cool, looks like my mention showed up but not my like! https://aaronparecki.com/2021/10/04/9/my-blog
# Loqi [Aaron Parecki] you know what's *not* down right now?
my blog ⬇
https://aaronparecki.com/
akevinhuang2, Seirdy0, Seirdy, Jeremy, [tw2113_Slack_] and edburns[d] joined the channel
# capjamesg[d] petermolnar How easy / secure is it to run your own email?
# 
petermolnar That is a deep and complex question: if you go with something that holds your hand, it's not too hard and secure enough, but there's no guarantee that the wonderful silo emails won't send you to spam.
# petermolnar if you start from scratch and by hand, it could take a while.
# petermolnar there's https://www.iredmail.org/ and https://mailinabox.email/ for simple-and-good enough
# petermolnar there's postfix + opendkim + opendmarc + dovecot + spamassassin + postgrey + clamav for all by yourself
# petermolnar and there's https://www.cyrusimap.org/ + https://www.exim.org/ for the cathedral experience (as in https://en.wikipedia.org/wiki/The_Cathedral_and_the_Bazaar )
# 
Allie ran their own email for many years
# Allie it's relatively painless unless you want to email people hosted by microsoft
[snarfed] joined the channel
# petermolnar ... or gmail with a fresh IP, or yahoo on random occasions, or mail.ru for whatever reason.
# petermolnar looks at his homeserver doing so
# petermolnar is getting a "business" internet connection an option for that?
# petermolnar I have a Virgin Media Business at home; before that, Zen(.co.uk), both worked flawlessly
# [snarfed] ...buuuuut if you want to own your email for privacy, data sovereignty etc reasons, consider something like https://thehelm.com/ instead
# petermolnar wait, there was something new as well, not just helm
# petermolnar yep, go with this one
gRegor joined the channel
# [snarfed] unrelated, capjamesg inspired me to put together the regexp queries for feed types in Indie Map. counts by link rel + type are https://snarfed.org/indiemap_link_rel_type_counts.csv.txt , complete list of sites by rel + type is https://snarfed.org/indiemap_link_rel_type_per_site.csv.txt
tetov-irc joined the channel
