Ticket created by ~xenrox on ~sircmpwn/lists.sr.ht
The UpdatePatchset mutation currently fails if one wants to update a patch from a list where the user does have
moderate
access but isn't the list owner. This is caused by this query which hard-codes the list owner as the current user.Instead the access table should be consulted.
Comment by ~xenrox on ~emersion/hut
There was a small discussion in IRC on how this could look:
<emersion> it's a bit tricky to describe in CLI flags though… <emersion> maybe --files <glob> --cache-control <value> <ddevault> --cache 'foo/*.html:cache-control header goes here'? <emersion> or that <ddevault> alternatively, --cache=cache.rules where cache.rules is a file with slightly more structure
I would be in favor of the last option. A file should allow the biggest flexibility and leaves room for future settings without adding too many flags. The file could look similar to this - if we want to avoid pulling in new dependencies and use scfg:
notFound "404.html" fileConfigs { glob "*.png" { cacheControl "max-age=100" } glob "*.html" { cacheControl "max-age=1" } }
With that we could add a
site-config
flag to hut and deprecatenot-found
(or we leave notFound out from the file). Do you have further thoughts on this ~emersion?
Sorry for the late reply. This looks perfect to me! Thank you very much! “~xenrox” outgoing@sr.ht writes:
Does something along the lines of this patch work for you?
Comment by ~xenrox on ~emersion/hut
Does something along the lines of this patch work for you?
Comment by ~xenrox on ~emersion/hut
I have created a simple example repository, that uses a build manifest with the hut development version to set the custom README. Maybe this is enough of a reference point for now.
Comment by ~xenrox on ~emersion/hut
hut can already do this. It was added in this commit, which is not released yet. That is why I haven't yet send a patch to update man.sr.ht.
Comment by ~xenrox on ~emersion/gqlclient
The time documentation recommends that it should typically not be passed as a pointer and that IsZero can be used to determine uninitialized times. But if there are indeed legitimate use-cases where you would use the zero value, then time should probably be removed from this case. Otherwise the backend has to check with
IsZero
.
Ticket created by ~xenrox on ~emersion/gqlclient
Some initial discussion happened in hut-dev
Ticket created by ~xenrox on ~emersion/gqlclient
omitempty
does not work ontime.Time
structs.See https://lists.sr.ht/~emersion/hut-dev/%3C20220218135156.614905-1-contact%40emersion.fr%3E