Dmitry Marakasov @AMDmi3@octodon.social

@feld
So you basically can make life hell for anyone in Hungary just by sending them encrypted emails? Cool, but still hard to believe.

Repology finally has atom feeds, so if you are maintainer of some packages in one of 180+ repositories it supports including most #linux and #bsd distros, it will now report when your packages need attention.

https://octodon.social/media/zZOYQtlJvWiUvQgR44o

@fdroidorg
You seriously need some more QA

Repology has support for history of version updates for all projects now. It's interesting to see how new software releases spread along the repositories (https://repology.org/metapackage/subversion/history for example).

Next step is per-maintainer histories and RSS/Atom feeds!

The aggregated history of ALL software releases is possible now too, but I don't find it practically usable, as there are A LOT of them.

@gamingonlinux
Mine is #FreeBSD

Sometimes I think #GNU exists for the sole purpose of finding a use for their #guile.

Recently introduced #tor DOS protection seems to be working, after update to 0.3.3.3 my FreeBSD tor relay no longer sees spurious mbuf clusters usage spikes. Thanks for making tor more and more bulletproof!

@Lodart
Can't decide either. I lean towards ML as it has boost/fav counters displayed

I need an android app which allows doing something useful. Coz I'm tired of chatting, gaming or reading garbage from social media. Something:

- f/oss only
- plug and play (no need to manually search for a project to work on as with e.g. stringlate)
- simple (no need to edit huge texts, as it's a phone)
- developer oriented

Small string translations, code reviews, bug triaging, image classification, proofreading maybe.

Any suggestions?

@debugninja
A lot of people has already left, so it's really questionable whether it has created a welcoming environment. What really hurts the community is connivance with people leaving the project.
@dewb

@saper because it's constant expression which evaluates to TRUE. Well the problem is explained in https://www.postgresql.org/message-id/2374.1366296417%40sss.pgh.pa.us

And between fiddling with the planner and introducing SQL query template engine I think I'll pick the latter, as it's at least straightforward and predictable.

@saper the extra table scan is bad, while that table is not really used in the request. Planner is not able to deduce that and avoid the scan, and there's no way to make join or subrequest conditional, so I have to reside to dynamicaly generating SQL queries, which I've hoped to avoid

@saper plain UNION might work. The idea is that I have a lot of joined tables each of which serves specific part of large AND condition, and when that part is equivalent TRUE I'd like to avoid unneeded table scan.

#PostgreSQL has failed me for the first time. Being unable to optimize seemingly simple SELECT ... WHERE TRUE OR field IN (SELECT ...) and eliminate the join, it leaves me no choice but to shove template engine into my database handling code...

@smortex I'd mention all listed non-build-time licenses

LICENSE=MIT BSD#CLAUSE (look into files mentioned in "Runtime code" section for actual # of clauses) APACHE20 GPL?? (mentioned in "Class Library code", not sure of version and "or later")
LICENSE_COMB=multi
LICENSE_FILE=${WRKSRC}/LICENSE

Not many distros seem to set this correctly: https://repology.org/metapackage/mono/information#Licenses

@mpts Be sure to also check out https://wiki.freebsd.org/PortsLongtermTargets