Filesystem walks of large GOPATHs/module caches can take seconds,
especially on systems with slow filesystems like MacOS and WSL. We don't
want to block completion requests on walks finishing. At the same time,
cancelling a walk midway through results in an unusable cache, where we
don't know which parts have been scanned so far.
The best option is to run the walks in a separate goroutine. Then we can
detach and let them finish. On the other side, we need to be able to
reattach for the next completion request.
Introduce a new method on caches, ScanAndListen, which first processes
all the items in the cache, then notifies of any new items. This allows
us to reattach to an existing scan without missing anything.
The background scan introduces concurrency to the resolvers where there
wasn't any before. We can't use mutexes, because there's no way to stop
Lock() when a context expires. Use a 1-element semaphore channel to
accomplish the same effect.
Along the way: Only rescan GOPATH if the resolver has been cleared. None
of this makes sense for GOPATH without that. Fix a bug where we were
scanning the main module twice in module mode. Stop loading exports in
module tests, it slows them down a ton.
Change-Id: I978efae733ccba0c0cdc8e8fe6892bf5f15feac8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/213217
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now that we're storing module information per-directory, we don't need
to pre-compute the need for a replace statement. And we never have the
package name in the place it used to be set.
Change-Id: I3b0845dc49f52f8c449840410dbb786fe903d29d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212861
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We have multiple use cases for scanning: goimports, import completion,
and unimported completions. All three need slightly different features,
and the latter have very different performance considerations. Scanning
everything all at once and returning it was not good enough for them.
Instead, design the API as a series of callbacks for each
directory/package: first we discover its existence, then we load its
package name, then we load its exports. At each step the caller can
choose whether to proceed with the package. Import completion can stop
before loading exports, goimports can apply its directory name
heuristics, and in the future we'll be able to stop the scan short once
we've found all the results we want for completions.
I don't intend any significant changes here but there may be some little
ones around the edges.
Change-Id: I39c3aa08cc0e4793c280242c342770f62e101364
Reviewed-on: https://go-review.googlesource.com/c/tools/+/212631
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We intend to use the GOPATH resolver's results during LSP
autocompletion. That means we have to be able to cache its data, same as
we do for modules. Convert it to use a dirInfoCache.
Cache exports in the dirInfoCache. Along the way, store exports as slices
rather than maps. We don't need the extra structure the vast majority of
the time, and the memory overhead is nontrivial.
Change-Id: If267d6b00da2163a960b93b2cf2088ec2538f73d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/205162
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
To avoid proposing unusable completions, such as those in modules that
need a replace statement to be usable, we need to know what module a
directory is in. That involves walking up the directory tree to find a
go.mod file, which is expensive to do over and over. Really, we just
need to check if the directory we're in has a go.mod file, then use the
parent dir's results.
Add module information to the cache and use it when figuring out what
module a dir is in.
Change-Id: Ia74ba9b37d73fca5e6786a94c73c8fd71b591645
Reviewed-on: https://go-review.googlesource.com/c/tools/+/202541
Run-TryBot: Heschi Kreinick <heschi@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Since a user's module cache is generally going to be much bigger than
their main module, one would expect that caching just information about
the module cache would be sufficient. It turns out that's not correct.
When we discover something in the module cache, we have to make sure
that a different version of it isn't already in scope. Doing that can
require information about the main module or replace targets, so that
needs to be cached too.
Concretely, when I'm working in x/tools, if a scan discovers a version
of x/tools in the module cache, it should usually ignore that version.
But that might not be true in more complicated cases, especially those
involving nested modules whose boundaries change.
So, cache everything except GOROOT. Since the new data is mutable,
we store it separately from the module cache data so that it can be
discarded easily between runs.
Change-Id: I47364f6c0270fee03af8898fec6c85d1b9c8d780
Reviewed-on: https://go-review.googlesource.com/c/tools/+/202045
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Scan most sources, including GOPATH, the module cache, the main module,
and replace targets as appropriate. Use the cached stdlib instead of
scanning GOROOT.
We heavily cache the contents of the module cache, so performance is
decent. But we have to look at all the modules not in the module cache
too to get the right versions of modules (see
(*ModuleResolver).canonicalize), which currently isn't cached at all,
even just for a single run. That ends up being pretty expensive.
The implementation changes are relatively small; add package name
loading to scan(), cache that result, and allow callers to control what
directories are scanned so that it can skip GOROOT.
I also cleared out most of the stdlib from the unimported completion
test and added a simple external completion to it for safety's sake.
Change-Id: Id50fd4703b1126be35a000fe90719e19c3ab84bf
Reviewed-on: https://go-review.googlesource.com/c/tools/+/199178
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Packages found in the module cache do not change. When we encounter a
directory we have already processed in the module cache, skip that
directory and add the packages that have already been computed.
Change-Id: Ib1bf0bf22727110b8073b415b145034acceb6787
Reviewed-on: https://go-review.googlesource.com/c/tools/+/186921
Run-TryBot: Suzy Mueller <suzmue@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
To check if a package is in a module that is in scope, the module
resolver checks if there are Go files that would be included in a
package in the directory matching the import path in scope.
If this directory is in the module cache and we have saved it as a
package, we know this directory contains Go files, and do not have to
read the directory.
Change-Id: I7c9365ce42c760ab95bc68b036212120895c89fb
Reviewed-on: https://go-review.googlesource.com/c/tools/+/186922
Run-TryBot: Suzy Mueller <suzmue@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
The module cache can only be added to, so any information discovered
about directories that are within a module in the module cache will
not change. Store the information we have discovered about the module
cache.
Updates #32750
Change-Id: I56c88f03f6a364221198fb032b139208497cd0e5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/188762
Reviewed-by: Heschi Kreinick <heschi@google.com>
