resolves#286
this uses the upstream tar.
it also has support for changing ownership of the files unpacked (which the keter implementation also seemed to manage).
this also deletes the stack based ci in favor of a cabal based ci, it broke for some reason and I didn't want to play stack whackamole.
resolves https://github.com/snoyberg/keter/issues/286
this uses the upstream tar.
it also has support for changing ownership of the files unpacked (which the keter implementation also seemed to manage).
this also deletes the stack based ci in favor of a cabal based ci, it broke for some reason and I didn't want to play stack whackamole.
use upstream unpack
bump filepath
forM more
import forM
trash result
don't use a traverse but a fold
update changelog
add extra deps
clear stack
add cabal based action
drop windows support
clear stack based ci
Add note on changing to cabal ci
I just don't want to figure out why this broke.
bump keter
AWS LB TG healthcheck cannot currently be made to send any custom Host
header. It always sends the target's (keter's) IP address in there.
Together with vhosting, this makes it impossible to healthcheck
individual webapps running under keter.
Have a partial remedy with a /keter-health endpoint, that if enabled
always responds with status 200.
The fields of KeterConfig were being parsed from Yaml with a "flipped string constants" mistake:
* kconfigUnknownHostResponse -- from "missing-host-response-file"
* kconfigMissingHostResponse -- from "unknown-host-response-file"
Fix that; use a speck of typelevel spice to showcase "how not to" fall victim
to this class of mistakes again, if desired. The Tagged helper is:
* zero-cost;
* simple;
* easy to use;
* attaches a constant string to a type, and transparently tracks it
for as long as you allow it to -- e.g. in this case, from KeterConfig
data-decl all the way until readFile calls.
