sanin.dev
commits
Also adds the relevant DB helpers for this.
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
This hinders atprotation and is very appview-specific anyway.
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: ItsHarper <git-f4cbe@harperandrews.org>
this avoids stale raw content from being sent to clients.
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: dusk <y.bera003.06@protonmail.com>
Signed-off-by: dusk <y.bera003.06@protonmail.com>
Signed-off-by: dusk <y.bera003.06@protonmail.com>
Signed-off-by: dusk <y.bera003.06@protonmail.com>
this was broken by the change that reworked the ResolveRepo middleware.
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: ItsHarper <git-f4cbe@harperandrews.org>
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
as a result, this lets us avoid making another call to knots.
Signed-off-by: oppiliappan <me@oppi.li>
uses the newly added description filter that only autlinks, and renders
backticks, bold and italics.
Signed-off-by: oppiliappan <me@oppi.li>
this sanitizer only renders a tiny subset of markdown, to be used in
inline elements like PR titles and repo descriptions etc.
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: oppiliappan <me@oppi.li>
sqlite db still has `at_uri` column to easily use as reference
Signed-off-by: Seongmin Lee <boltlessengineer@proton.me>
Signed-off-by: Seongmin Lee <boltlessengineer@proton.me>
There is no reason for `ResolvedRepo` to not directly inherit `Repo`
Signed-off-by: Seongmin Lee <boltlessengineer@proton.me>
tailwind seems to apply this to every html element. when applying
font-mono to an element, it inherits the ss01 font feature. this results
in a different `a` characters from when inside `pre` and `code` blocks,
where it resets the list of font-features.
i can't seem to find a good fix for this issue. removing the ss01
feature altogether seems like a good fix, the number alternates in inter
aren't all that different in my opinion.
Signed-off-by: oppiliappan <me@oppi.li>
- remove redundant divs in base layout
- allow scrolling changed-files listing
- show more of the file in tree views
- unify repo-index and repo-tree views
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Replace `Issue.IssueAt` to `Issue.Rkey`. `issue_at` column still exist,
but not used and will be removed in future.
Remove `db.SetIssueAt()` in flavor of `Rkey`.
Create rkey *before* adding to DB, and save same rkey to PDS.
This is safe as `tx` is will be dropped when `ctx` is canceled with
error.
Signed-off-by: Seongmin Lee <boltlessengineer@proton.me>
Otherwise, when programs create files in $HOME, others that check the Git
repo's state may be sad (or at least emit warnings in response to) it
being dirty.
Signed-off-by: Winter <winter@winter.cafe>
IMO, we shouldn't make errors opt-in to actually show upon push, at least
not if/until we have a UI to show them to the user.
Signed-off-by: Winter <winter@winter.cafe>
This moves away from the old method of creating a container with some
shared volumes to one that most users would expect: any changes made
in one step will be accessible by the following steps, and not only if
they're in the workspace or `/etc/nix`. This also paves the way for a
more generic Docker image engine, as users can do things like `apt install`
without the results being blown away across steps.
Signed-off-by: Winter <winter@winter.cafe>
It's possible for SetupWorkflow to provision resources and also fail, so
this makes sense for robustness.
Signed-off-by: Winter <winter@winter.cafe>
Signed-off-by: Winter <winter@winter.cafe>
Signed-off-by: Winter <winter@winter.cafe>
Signed-off-by: Winter <winter@winter.cafe>
and fix empty-repo rendering.
Signed-off-by: oppiliappan <me@oppi.li>
From https://datatracker.ietf.org/doc/html/rfc9207#section-2.4 :
"Clients MUST [...] compare the result to the issuer identifier of the
authorization server where the authorization request was sent to. [...]
If the value does not match the expected issuer identifier, clients MUST
reject the authorization response and MUST NOT proceed with the
authorization grant."
Signed-off-by: Mike P <fentiger@mythik.co.uk>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
This reverts commits 1931b62baa4452616203ccbb356da72688bc4587 and 86d0b4060de074759eff40666b90be86c244ab23.
Signed-off-by: Winter <winter@winter.cafe>
Causes too many regressions, unfortunately:
- https://discord.com/channels/1361963801993285692/1361991850327674932/1406398439830323292
- https://discord.com/channels/1361963801993285692/1361991850327674932/1406401591627747519
Ideally, our templates wouldn't be whitespace-sensitive (example #1),
but let's stop the bleeding while we figure it out.
Signed-off-by: Winter <winter@winter.cafe>
Signed-off-by: Winter <winter@winter.cafe>
I guess it's not idempotent in some cases? Looks like it's happy with
these, though, as rerunning it after this commit produces no changes.
Hopefully this is a one time thing...
Signed-off-by: Winter <winter@winter.cafe>
Since Zed can't easily use our wrapped Prettier (see previous commit),
and we're now using three formatters, let's just add one command to
format all of our code correctly.
Signed-off-by: Winter <winter@winter.cafe>
This also removes the Zed Prettier configuration, because AFAICT, it
never actually worked, and will not work with how we install Prettier
now. :(
Signed-off-by: Winter <winter@winter.cafe>
This does result in the removal of some options that aren't supported by
the SQLite driver [0], but IMO we can stick with the defaults for these.
[0]: https://github.com/mattn/go-sqlite3#connection-string
Signed-off-by: Winter <winter@winter.cafe>
`sql.DB` pools connections, which means that there's a good chance that
our foreign keys aren't actually being enforced, unless a goroutine
happens to be using the connection that was made when we set `pragma
foreign_keys = 1`.
Signed-off-by: Winter <winter@winter.cafe>
do not return errors from ingesters, this causes the read loop to be
killed.
Signed-off-by: oppiliappan <me@oppi.li>
Signed-off-by: oppiliappan <me@oppi.li>
Also adds the relevant DB helpers for this.
Signed-off-by: Anirudh Oppiliappan <anirudh@tangled.sh>
Signed-off-by: oppiliappan <me@oppi.li>
tailwind seems to apply this to every html element. when applying
font-mono to an element, it inherits the ss01 font feature. this results
in a different `a` characters from when inside `pre` and `code` blocks,
where it resets the list of font-features.
i can't seem to find a good fix for this issue. removing the ss01
feature altogether seems like a good fix, the number alternates in inter
aren't all that different in my opinion.
Signed-off-by: oppiliappan <me@oppi.li>
Replace `Issue.IssueAt` to `Issue.Rkey`. `issue_at` column still exist,
but not used and will be removed in future.
Remove `db.SetIssueAt()` in flavor of `Rkey`.
Create rkey *before* adding to DB, and save same rkey to PDS.
This is safe as `tx` is will be dropped when `ctx` is canceled with
error.
Signed-off-by: Seongmin Lee <boltlessengineer@proton.me>
This moves away from the old method of creating a container with some
shared volumes to one that most users would expect: any changes made
in one step will be accessible by the following steps, and not only if
they're in the workspace or `/etc/nix`. This also paves the way for a
more generic Docker image engine, as users can do things like `apt install`
without the results being blown away across steps.
Signed-off-by: Winter <winter@winter.cafe>
and fix empty-repo rendering.
Signed-off-by: oppiliappan <me@oppi.li>
From https://datatracker.ietf.org/doc/html/rfc9207#section-2.4 :
"Clients MUST [...] compare the result to the issuer identifier of the
authorization server where the authorization request was sent to. [...]
If the value does not match the expected issuer identifier, clients MUST
reject the authorization response and MUST NOT proceed with the
authorization grant."
Signed-off-by: Mike P <fentiger@mythik.co.uk>
Causes too many regressions, unfortunately:
- https://discord.com/channels/1361963801993285692/1361991850327674932/1406398439830323292
- https://discord.com/channels/1361963801993285692/1361991850327674932/1406401591627747519
Ideally, our templates wouldn't be whitespace-sensitive (example #1),
but let's stop the bleeding while we figure it out.
Signed-off-by: Winter <winter@winter.cafe>