The fonts we embed in Zed binary (Zed Sans & Zed Mono) weigh about 30Mb in total and we are cloning them several times during startup and loading of embedded assets (once explicitly in Zed and then under the hood in font-kit). Moreover, after loading we have at least 2 copies of each font in our program; one in .rdata and the other on the heap for use by font-kit.
This commit does away with that distinction (we're no longer allocating the font data) and slightly relaxes the interface of `TextSystem::add_fonts` by expecting one to pass `Cow<[u8]>` instead of `Arc<Vec<u8>>`. Additionally, `AssetSource::get` now returns `Cow<'static, [u8]>` instead of `Cow<'self, [u8]>`; all existing implementations conform with that change.
Note that this optimization takes effect only in Release builds, as the library we use for asset embedding - rust-embed - embeds the assets only in Release mode and in Dev builds it simply loads data from disk. Thus it returns `Cow<[u8]>` in it's interface. Therefore, we still copy that memory around in Dev builds, but that's not really an issue.
This patch makes no assumptions about the build profile we're running under, that's just an intrinsic property of rust-embed.
Tl;dr: this should shave off about 30Mb of memory usage and a fair chunk (~30ms) of startup time.
Release Notes:
- Improved startup time and memory usage.
This PR changes the theme loading to use the JSON themes bundled with
the binary rather then the Rust theme definitions.
### Performance
I profiled this using `cargo run --release` to see what the speed
differences would be now that we're deserializing JSON:
**Before:** `ThemeRegistry::load_user_themes` took 16.656666ms
**After:** `ThemeRegistry::load_user_themes` took 18.784875ms
It's slightly slower, but not by much. There is probably some work we
could do here to bring down the theme loading time in general.
Release Notes:
- N/A
Although we now usually get the .ips file alongside a panic, when we
don't it's much harder to track down where panics happen because we're
not showing inlined frames.
Try iterating over all symbols in a frame to see if that improves panic
reporting.
This uploads apples crash reports to our servers when telemetry is
enabled.
Rather than jumping via the website, it uploads to collab directly. (I'd
like
to update the panic handler to do this too to make it possible to work
on that end-to-end without zed.dev running).
Release Notes:
- Added reporting of crashes when telemetry is enabled
This essentially shaves off about 10% off of an incremental build after
project change and potentially more if you're changing stuff like
`welcome` that's very close to the `zed` crate in the dep graph. That's
because macro expansion takes place even in incremental builds it seems?
And zed (lib) + zed (bin) could take up to 4 seconds out of an
incremental build, which is a *lot* in a 10s build. In reality though it
shaves 1 second off of 5 seconds incremental 'welcome'/ 1s off of 10s
'project' builds.
Note that we had `assets` crate in the past (removed in #2575 /cc
@maxbrunsfeld), but this is a bit different, because `assets` is a
dependency of *just* zed and nothing else. We essentially cache macro
expansion results ourselves.
Release Notes:
- N/A
The client sends events to our end point, and the endpoint is what determines what analytics database is used to store the data. The client should be generic and not mention the name of the service being proxied to through our server.
If we find a previous installation_id, then we send `open`. If we don't find a previous installation_id, then we sent as `first open`. If we fail, we mark it as `open` so that we don't accidentally bloat our `first open` stats.
I just panicked and wanted to see the cause, but forgot that panic files
get deleted when Zed uploads them.
Release Notes:
- Panics are now written to `~/Library/Logs/Zed/Zed.log`
