This reverts commit bba678e5c5.
Reason for revert: didn't fix the bug and created new ones (fj#794)
Change-Id: I0450205d3041b6c876737151a4051081c1366f1d
we generally do not want to catch or throw these. catching them to print
and discard is fine, tests are largely exempt, and cases in which we can
be certain where the exception came from are also fine to *catch*. we'll
try to never *throw* (or rethrow) these if possible though because doing
so will make it impossible to construct async traces for the exceptions.
Change-Id: I3b71c32ecd16afc2246c946472f5629a1fa31f2c
This seems to be a copypasta mistake that slipped through when writing the
initial tests. Changes the ValuePrintingTests::tPath to actually test a path
value instead of a string.
Change-Id: I20fe72635a0b1a6a0d72f674a4619da0de9f3d44
Well, I was trying to figure out
https://git.lix.systems/lix-project/lix/issues/729 in which this feature
is clearly just broken on macOS, but frustratingly, it seems that it
*does* work, except for the daemon. um........ sure.
Change-Id: Iaa962f045c16fdfa82854151c90a03e5cf0eea47
Yikes!! I wonder if we have any other ones of these that just .. didn't
get added to a meson file?
Change-Id: I19480ab03cdbecf608e523d5b6c3980233f4f445
or more accurately, wrap them in a nix::Error subclass so we can display
them properly without crashing, and add some error context if available.
fixes#642fixes#753fixes#759fixes#769
Change-Id: I1aad0c0501fea83f9de3a1335eaa6adc20721616
technically it doesn't *have* to be NeverAsync, but not marking it as
such unconditionally requires templating DebugState over asyncness of
its callback (which then requires templating EvalState, which, *NO*.)
Change-Id: I4980d45b541c2e40328beac139b18c6c1ba0957c
allow opting in to serialization as integers via a trait type instead,
and add string-list serializers for the feature flag set enumerations.
fixes#738
Change-Id: I2746eb5ef1f15c01b4e681f9ba1615b6c6e64f44
we want to own this specialization fully so we can change the default
serializer behavior without also forcing downstream users of our code
to use the same behavior. it'll also let us do things we cannot do in
regular nlohmann::json, such as selectively enabling serialization of
enums as integral types, or using `to_json`/`from_json` overloads for
not-default-constructible types instead of serializer specializations
Change-Id: I91a1db362e37d654090f1824b1cd3ce783d32134
this will become a proper specialization of `nlohmann::basic_json` soon.
specialing basic_json will let us get rid of our `adl_serializer` hacks,
and it'll open the door to better enum serializing behavior without also
forcing all those who use lix as a library to set certain defines (which
may not even be possible depending on how those users use json already).
Change-Id: I5228d2b9df581a189552c993363207cfbd20f445
this doesn't do much, just wrap a few nlohmann headers in headers of our
own (and delete includes we don't need because they're transitively seen
by other includes). doing this now will make the next change much nicer.
Change-Id: I166933102ea86bb5322ebbf9ba9411f96032a53b
if a copyNAR generator was not drained to completion it would not read
the full nar data from its source. this could happen if the copier was
passed to parseAndDump wrapped as a source because copyNAR would yield
nar metadata *before* it had read it, and GeneratorSource will drain a
generator fully *only* if the source is allowed to throw EndOfFile. in
the parseAndDump case this never happened because parseAndDump expects
to be given an unterminated stream, and thus the combination left some
nar metadata in the input Source, breaking the remote store protocols.
fixes#732
Change-Id: Ia59a53375992bfcdb7bc6b37764ca779622bc8f7
well, oops. on slow io (as can happen with ssh remote builders) we could
have extended a nar read buffer past what was actually read, injecting a
span of zeroes into the read buffer where we requested some data but got
a partial result instead. also add some tests that would've caught this.
Change-Id: I67aa06b4715aeec6a5bdacaafa9b79849e664e2f
there's no real usefulness to this output in build logs. in interactive
development builds -v can be passed to meson to restore the old output.
Change-Id: Iac4c0573fd1cac5fc9044b950ef52ee50448e6fc
connection pools of remote stores can currently block. this is not a
problem when each request runs on a dedicated thread, but with async
code this is no longer true. if a remote store has exhausted all its
available connections on one executor and another job starts *on the
same executor* we'll deadlock if that job makes another remote store
request. unlike with sqlite previously it's not reasonable, not even
necessary, to make the pools unbounded: since we use pools only with
remote stores we can asyncify all of them at once, and since they're
leaves of all call stacks we do not have much code to change either.
Change-Id: I8c457e27893e22c2cfc35933307a5283c986805a
sadly this is a visitor-only interface; async generators are not yet a
thing and preliminary benchmarks say that overhead would be too large.
Change-Id: I0460d18eba94441cf3101d46bfffdb69cdda81d1
we'll use this in NarAccessor to provide actually safe indexing of
archives. NarAccessor currently is not fully correct: it relies on
the parser not buffering anything to produce correct file offsets,
but only the nar implementation itself can reasonably expect that.
Change-Id: I64f300b86d8844b876a3ace723546ea7d7b4628b
this is where they should've been from the start, but during the first
rewrite it made little sense to move them. we have bigger plans today,
so we'll finally clean that up too. note the `Map` transform type that
is needed to make the current macros work. it shall be only temporary.
Change-Id: I928d197dbfe27b68cf8634d149c3259e86fbf123
This is kind of a painful change, because a lot of code grew around the
bad abstraction, but the goal here is to abstract formals in a way that
allows adding other means of pattern matching / argument destructuring
in the future.
Change-Id: If4e681a4be3d1f42ceea81a8e07297f0d08acc80
also mark the sync version as NeverAsync. a blocking wait on sqlite
locks in a coroutine may never finish if it's a different coroutine
on the same executor that is holding the lock, not another process.
this propagates to the sqlite core interface, but no further. we'll
assume that caches do not block on a database for very long, and we
can't reasonably propagate never-async-ness out of stores unless we
touch everything we'd touch for the async transition, again, twice.
store code already assumes that it can block for however long it'll
feel like that moment. we keep thread pools around for this reason.
Change-Id: I62f77e1ac333cbe2e4e646dbcb1571463f2cf3fc
4138fc7622 mistakenly removed an early
exit from non-main worker threads. this led to exceptions not ending
`process()` calls in a timely manner and instead draining the entire
work queue first, which for e.g. Interrupted errors would cause many
duplicated reports per worker thread instead of only one per thread.
it also did not properly rethrow a work item exception in all cases,
e.g. when all work had completed by the time `process()` was called.
due to a mistake in the thread starting check it was possible that a
system would require n² work items to start n worker threads if some
work items process quickly enough while other items block for a bit.
Change-Id: I7d59483580cda1c19980f9296074216a0fbf2c4e
We will link to boehm inside of the test suite if it depends on libexpr
but we don't need it for other tests.
For anything linking to libexpr internally, just like externally, it
needs to link to boehm.
Likewise with kj and libutil (or really any lix), so we should just make
it automatic.
Change-Id: I2bb9ec4668e6ff741b4139fdce167f278eb71c7e
ifd requires building which uses async code today, fetching requires
network io which really should use async code but does not yet. eval
itself cannot be made async for performance reasons. offloading work
to other threads is possible but questionable as currently there are
no uses in which evaluation must take place from within asynchronous
operations, which are currently confined entirely to store behavior.
Change-Id: I1e3c86fca19b042d79332e9a5ef9546f90a6448b
There's a race condition where awaitData could early-return for data
coming from a 404 response or similar and thus not rethrow the exception
that is forthcoming, and a related race during transfer setup (which
could retry a transfer *twice* per retry round).
This would then cause substitution failures like below since the exception
isn't caught in HttpBinaryCacheStore::getFile as intended, but instead
by an exception handler downstream of `drain()` which would error out
the entire operation.
Symptom:
» nix-build ./docs-service.nix -o "docs-service-result"
error: unable to download 'https://cache.nixos.org/7mr3fy8w66gi5inmf0jkkkl90lxy4jyg.narinfo': HTTP e
rror 404 ()
response body:
This is kind of a hack in how it is implemented: it assumes that you
can't intentionally be receiving a large unsuccessful response since in
such a case, `awaitData` will wait for finish() to be called to throw an
exception and will never escape until the download finishes, while
continuing to buffer the entire response into memory, which could be bad
if an error response had a large payload.
That said, nobody is sending Lix 1GiB of 404, so meh I guess, and this
is how it is seemingly intended to work. That was a design flaw of the
thing before any of the Lix team got our paws on it.
I tested this by adding _exit(0) inside the expected exception catch and
then running the offending command repeatedly to see if the symptom ever
appeared again, and it did not.
Needs cherry-pick to 2.92 and a 2.92.1 release once reviewed.
Fixes: https://git.lix.systems/lix-project/lix/issues/635
Co-Authored-By: lix@jade.fyi
Change-Id: If54f6eeaad60b5ca9d5b77d4d9232da1d295e7d1
The code for serialization Expr nodes back into (pseudo-)Nix has been
removed for being subtly error-prone and tedious to maintain. Instead,
`nix-instantiate --parse` now prints a JSON representation of the AST.
Usage patterns of the --parse flag I've found in the wild:
1. Check if a file is well-formed, i.e. discard output and test exit code
2. Get parser errors from a file, i.e. discard stdout and use stderr
3. Nixfmt uses --parse to test equivalence pre/post format, and that property is (should be?) preserved
None of these should break with the current change
Closes#487
Change-Id: Icdbaad17790f2ad8765fa08e02e6597ee4c7a909
references remain forbidden because std::optional does not want to
contain them, and specializing generators to use pointers where we
can't use optionals is simply too much work for a feature we don't
even need. reference wrappers and bindings still work well enough.
Change-Id: I2e6ca74719584ce16e2357c452fdd5c5a9e23d5a
