Commit graph

59 commits

Author SHA1 Message Date
pennae a385e51a08 rename SymbolIdx -> Symbol, Symbol -> SymbolStr
after #6218 `Symbol` no longer confers a uniqueness invariant on the
string it wraps, it is now possible to create multiple symbols that
compare equal but whose string contents have different addresses. this
guarantee is now only provided by `SymbolIdx`, leaving `Symbol` only as
a string wrapper that knows about the intricacies of how symbols need to
be formatted for output.

this change renames `SymbolIdx` to `Symbol` to restore the previous
semantics of `Symbol` to that name. we also keep the wrapper type and
rename it to `SymbolStr` instead of returning plain strings from lookups
into the symbol table because symbols are formatted for output in many
places. theoretically we do not need `SymbolStr`, only a function that
formats a string for output as a symbol, but having to wrap every symbol
that appears in a message into eg `formatSymbol()` is error-prone and
inconvient.
2022-04-25 15:37:01 +02:00
pennae 8775be3393 store Symbols in a table as well, like positions
this slightly increases the amount of memory used for any given symbol, but this
increase is more than made up for if the symbol is referenced more than once in
the EvalState that holds it. on average every symbol should be referenced at
least twice (once to introduce a binding, once to use it), so we expect no
increase in memory on average.

symbol tables are limited to 2³² entries like position tables, and similar
arguments apply to why overflow is not likely: 2³² symbols would require as many
string instances (at 24 bytes each) and map entries (at 24 bytes or more each,
assuming that the map holds on average at most one item per bucket as the docs
say). a full symbol table would require at least 192GB of memory just for
symbols, which is well out of reach. (an ofborg eval of nixpks today creates
less than a million symbols!)
2022-04-21 21:56:31 +02:00
pennae 6526d1676b replace most Pos objects/ptrs with indexes into a position table
Pos objects are somewhat wasteful as they duplicate the origin file name and
input type for each object. on files that produce more than one Pos when parsed
this a sizeable waste of memory (one pointer per Pos). the same goes for
ptr<Pos> on 64 bit machines: parsing enough source to require 8 bytes to locate
a position would need at least 8GB of input and 64GB of expression memory. it's
not likely that we'll hit that any time soon, so we can use a uint32_t index to
locate positions instead.
2022-04-21 21:46:06 +02:00
Eelco Dolstra a0259a21a4 Don't hide repeated values while generating manifest.nix
Fixes #6243.
2022-03-22 13:18:56 +01:00
John Ericson 4d6a3806d2 Decode string context straight to using StorePaths
I gather decoding happens on demand, so I hope don't think this should
have any perf implications one way or the other.
2022-03-18 15:36:11 +00:00
John Ericson 91adfb8894 Create some type aliases for string Contexts 2022-03-11 22:30:10 +00:00
Eelco Dolstra e9c04c3351 Be more aggressive in hiding repeated values
We now memoize on Bindings / list element vectors rather than Values,
so that e.g. two Values that point to the same Bindings will be
printed only once.
2022-03-03 13:33:34 +01:00
Eelco Dolstra ecff9d969a printValue(): Don't show repeated values
Fixes #6157.
2022-03-03 13:18:23 +01:00
Eelco Dolstra 36c7b12f33 Remove std::string alias 2022-02-21 16:37:25 +01:00
Eelco Dolstra bd383d1b6f Make most calls to determinePos() lazy 2022-02-04 00:33:21 +01:00
pennae 5838354d34 optimize ExprConcatStrings::eval
constructing an ostringstream for non-string concats (like integer addition) is
a small constant cost that we can avoid. for string concats we can keep all the
string temporaries we get from coerceToString and concatenate them in one go,
which saves a lot of intermediate temporaries and copies in ostringstream. we
can also avoid copying the concatenated string again by directly allocating it
in GC memory and moving ownership of the concatenated string into the target
value.

saves about 2% on system eval.

before:

  Benchmark 1: nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
    Time (mean ± σ):      2.837 s ±  0.031 s    [User: 2.562 s, System: 0.191 s]
    Range (min … max):    2.796 s …  2.892 s    20 runs

after:

  Benchmark 1: nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
    Time (mean ± σ):      2.790 s ±  0.035 s    [User: 2.532 s, System: 0.187 s]
    Range (min … max):    2.722 s …  2.836 s    20 runs
2022-01-12 10:07:21 +01:00
Eelco Dolstra ca5baf2392 Turn mkString(Symbol) into a method 2022-01-04 19:09:40 +01:00
Eelco Dolstra ed93aec3c3 Remove non-method mkPath() 2022-01-04 18:45:16 +01:00
Eelco Dolstra 263a8d293c Remove non-method mk<X> functions 2022-01-04 18:40:39 +01:00
Eelco Dolstra cc08364315 Remove non-method mkString() 2022-01-04 18:24:42 +01:00
Eelco Dolstra 6d9a6d2cc3 Ensure that attrsets are sorted
Previously you had to remember to call value->attrs->sort() after
populating value->attrs. Now there is a BindingsBuilder helper that
wraps Bindings and ensures that sort() is called before you can use
it.
2022-01-04 18:00:33 +01:00
Silvan Mosberger 90700736c7 Introduce builtins.groupBy primop
This function is very useful in nixpkgs, but its implementation in Nix
itself is rather slow due to it requiring a lot of attribute set and
list appends.
2021-12-02 21:54:51 +01:00
Eelco Dolstra b6c8e57056 Support range-based for loop over list values 2021-11-25 16:31:39 +01:00
Kevin Amado d0e9e18489
toXML: display errors position
- This change applies to builtins.toXML and inner workings
- Proof of concept:
  ```nix
  let e = builtins.toXML e; in e
  ```
- Before:
  ```
  $ nix-instantiate --eval poc.nix
  error: infinite recursion encountered
  ```
- After:
  ```
  $ nix-instantiate --eval poc.nix
  error: infinite recursion encountered

       at /data/github/kamadorueda/nix/poc.nix:1:9:

            1| let e = builtins.toXML e; in e
             |
  ```
2021-11-13 20:33:34 -05:00
Maximilian Bosch 7c76964daa
libexpr: misc improvements for proper error position
When working on some more complex Nix code, there are sometimes rather
unhelpful or misleading error messages, especially if coerce-errors are
thrown.

This patch is a first steps towards improving that. I'm happy to file
more changes after that, but I'd like to gather some feedback first.

To summarize, this patch does the following things:

* Attrsets (a.k.a. `Bindings` in `libexpr`) now have a `Pos`. This is
  helpful e.g. to identify which attribute-set in `listToAttrs` is
  invalid.

* The `Value`-struct has a new method named `determinePos` which tries
  to guess the position of a value and falls back to a default if that's
  not possible.

  This can be used to provide better messages if a coercion fails.

* The new `determinePos`-API is used by `builtins.concatMap` now. With
  that change, Nix shows the exact position in the error where a wrong
  value was returned by the lambda.

  To make sure it's still obvious that `concatMap` is the problem,
  another stack-frame was added.

* The changes described above can be added to every other `primop`, but
  first I'd like to get some feedback about the overall approach.
2021-04-13 23:12:38 +02:00
Silvan Mosberger b70d22baca
Replace Value type setters with mk* functions
Move clearValue inside Value

mkInt instead of setInt

mkBool instead of setBool

mkString instead of setString

mkPath instead of setPath

mkNull instead of setNull

mkAttrs instead of setAttrs

mkList instead of setList*

mkThunk instead of setThunk

mkApp instead of setApp

mkLambda instead of setLambda

mkBlackhole instead of setBlackhole

mkPrimOp instead of setPrimOp

mkPrimOpApp instead of setPrimOpApp

mkExternal instead of setExternal

mkFloat instead of setFloat

Add note that the static mk* function should be removed eventually
2020-12-18 21:48:22 +01:00
Silvan Mosberger 12e65078ef
Rename Value::normalType() -> Value::type() 2020-12-17 14:45:45 +01:00
Silvan Mosberger d67e02919c
Rename ValueType -> InternalType, NormalType -> ValueType
And Value::type to Value::internalType, such that type() can be used in
the next commit to get the new ValueType
2020-12-17 14:45:22 +01:00
Silvan Mosberger 730b152b19
Make Value::type private
This is an implementation detail and shouldn't be used. Use normalType()
and the various is<Type> functions instead
2020-12-12 03:31:52 +01:00
Silvan Mosberger bf98903967
Add ValueType checking functions for types that have the same NormalType 2020-12-12 03:31:50 +01:00
Silvan Mosberger 9f056f7afd
Introduce Value type setters and make use of them 2020-12-12 03:31:48 +01:00
Silvan Mosberger fa307875e9
Introduce NormalType for the normal type of a Value
This will be useful to abstract over the ValueType implementation
details

Make use of it already to replace the showType(ValueType) function
2020-12-12 03:31:46 +01:00
Eelco Dolstra 50f13b06fb EvalCache: Store string contexts 2020-06-29 19:08:37 +02:00
Eelco Dolstra f89349f07e Merge remote-tracking branch 'origin/master' into flakes 2020-04-16 18:33:10 +02:00
Eelco Dolstra 10e17eaa58 ValueMap, VectorVector: Use traceable_allocator
We want to *trace* the 'Value *' arrays, not garbage-collect them!
Otherwise the vectors/maps can end up pointing to nowhere.

Fixes #3377. Closes #3384.
2020-04-16 17:30:13 +02:00
Eelco Dolstra b3e5eea4a9 Add function to allocate a Value in traceable memory 2020-04-16 17:30:05 +02:00
Eelco Dolstra c3c23a52ee Merge remote-tracking branch 'origin/master' into flakes 2019-12-04 00:31:09 +01:00
Eelco Dolstra 2d6f1ddbb5
Remove builtins.valueSize
Fixes #3246.
2019-11-28 13:52:42 +01:00
Peter Kolloch 2ba9f22715
De-duplicate struct PrimOp forward declaration 2019-11-10 10:02:22 +01:00
Eelco Dolstra f97d3753a1
Require flake.nix to be an attrset (not a non-trivial thunk) 2019-09-09 17:34:38 +02:00
aszlig 0ad643ed5c
libexpr: Use int64_t for NixInt
Using a 64bit integer on 32bit systems will come with a bit of a
performance overhead, but given that Nix doesn't use a lot of integers
compared to other types, I think the overhead is negligible also
considering that 32bit systems are in decline.

The biggest advantage however is that when we use a consistent integer
size across all platforms it's less likely that we miss things that we
break due to that. One example would be:

https://github.com/NixOS/nixpkgs/pull/44233

On Hydra it will evaluate, because the evaluator runs on a 64bit
machine, but when evaluating the same on a 32bit machine it will fail,
so using 64bit integers should make that consistent.

While the change of the type in value.hh is rather easy to do, we have a
few more options available for doing the conversion in the lexer:

  * Via an #ifdef on the architecture and using strtol() or strtoll()
    accordingly depending on which architecture we are. For the #ifdef
    we would need another AX_COMPILE_CHECK_SIZEOF in configure.ac.
  * Using istringstream, which would involve copying the value.
  * As we're already using boost, lexical_cast might be a good idea.

Spoiler: I went for the latter, first of all because lexical_cast does
have an overload for const char* and second of all, because it doesn't
involve copying around the input string. Also, because istringstream
seems to come with a bigger overhead than boost::lexical_cast:

https://www.boost.org/doc/libs/release/doc/html/boost_lexical_cast/performance.html

The first method (still using strtol/strtoll) also wasn't something I
pursued further, because it is also locale-aware which I doubt is what
we want, given that the regex for int is [0-9]+.

Signed-off-by: aszlig <aszlig@nix.build>
Fixes: #2339
2018-08-29 01:05:52 +02:00
Eelco Dolstra a92ed973e5
Store floating point numbers in double precision
Even on 32-bit systems, Value has enough space to hold a double.
2018-07-03 18:39:32 +02:00
Eelco Dolstra 53ec5ac69f
Fix some random -Wconversion warnings 2018-05-02 13:56:34 +02:00
Eelco Dolstra bcecc99007
Restructure installables handling in the "nix" command 2017-04-25 11:20:37 +02:00
Daiderd Jordan 023217f07c
use std::tuple for ValueMap allocator 2017-03-24 23:05:49 +01:00
Tuomas Tynkkynen 2cd468874f Include config.h implicitly with '-include config.h' in CFLAGS
Because config.h can #define things like _FILE_OFFSET_BITS=64 and not
every compilation unit includes config.h, we currently compile half of
Nix with _FILE_OFFSET_BITS=64 and other half with _FILE_OFFSET_BITS
unset. This causes major havoc with the Settings class on e.g. 32-bit ARM,
where different compilation units disagree with the struct layout.

E.g.:

diff --git a/src/libstore/globals.cc b/src/libstore/globals.cc
@@ -166,6 +166,8 @@ void Settings::update()
     _get(useSubstitutes, "build-use-substitutes");
+    fprintf(stderr, "at Settings::update(): &useSubstitutes = %p\n", &nix::settings.useSubstitutes);
     _get(buildUsersGroup, "build-users-group");
diff --git a/src/libstore/remote-store.cc b/src/libstore/remote-store.cc
+++ b/src/libstore/remote-store.cc
@@ -138,6 +138,8 @@ void RemoteStore::initConnection(Connection & conn)
 void RemoteStore::setOptions(Connection & conn)
 {
+    fprintf(stderr, "at RemoteStore::setOptions(): &useSubstitutes = %p\n", &nix::settings.useSubstitutes);
     conn.to << wopSetOptions

Gave me:

at Settings::update(): &useSubstitutes = 0xb6e5c5cb
at RemoteStore::setOptions(): &useSubstitutes = 0xb6e5c5c7

That was not a fun one to debug!
2017-02-08 21:51:02 +02:00
Eelco Dolstra 92d917b71a Fix GC build 2016-08-30 13:12:12 +02:00
Eelco Dolstra 26d92017d3 Add builtin function "partition"
The implementation of "partition" in Nixpkgs is O(n^2) (because of the
use of ++), and for some reason was causing stack overflows in
multi-threaded evaluation (not sure why).

This reduces "nix-env -qa --drv-path" runtime by 0.197s and memory
usage by 298 MiB (in non-Boehm mode).
2016-08-29 19:36:54 +02:00
Eelco Dolstra c0a7b84748 nix path-info: Add --json flag
Also, factor out JSON generation from value-to-json.{cc,hh}, and
support producing indented JSON.
2016-08-29 17:29:24 +02:00
Christian Theune 934642155c @eelco's feedback: downgrade to regular float for size, remove unused function. 2016-01-06 08:25:58 +01:00
Christian Theune 14ebde5289 First hit at providing support for floats in the language. 2016-01-05 00:40:40 +01:00
Eelco Dolstra b83801f8b3 Optimize small lists
The value pointers of lists with 1 or 2 elements are now stored in the
list value itself. In particular, this makes the "concatMap (x: if
cond then [(f x)] else [])" idiom cheaper.
2015-07-23 22:05:09 +02:00
Shea Levy 608110804c Make all ExternalValueBase functions const 2014-12-02 10:27:10 -05:00
Shea Levy 320659b0cd Allow external code using libnixexpr to add types
Code that links to libnixexpr (e.g. plugins loaded with importNative, or
nix-exec) may want to provide custom value types and operations on
values of those types. For example, nix-exec is currently using sets
where a custom IO value type would be more appropriate. This commit
provides a generic hook for such types in the form of tExternal and the
ExternalBase virtual class, which contains all functions necessary for
libnixexpr's type-polymorphic functions (e.g. `showType`) to be
implemented.
2014-12-02 10:27:04 -05:00
Eelco Dolstra 6bb4c0b712 mkList: Scrub better
Clearing v.app.right was not enough, because the length field of a
list only takes 32 bits, so the most significant 32 bits of v.app.left
(a.k.a. v.thunk.env) would remain. This could cause Boehm GC to
interpret it as a valid pointer.

This change reduces maximum RSS for evaluating the ‘tested’ job in
nixos/release-small.nix from 1.33 GiB to 0.80 GiB, and runtime by
about 8%.
2014-10-09 13:08:53 +02:00