lidavidm commented on code in PR #12:
URL: https://github.com/apache/arrow-nanoarrow/pull/12#discussion_r939160612
##
src/nanoarrow/nanoarrow.h:
##
@@ -261,6 +261,24 @@ ArrowErrorCode ArrowMetadataGetValue(const char* metadata,
const char* key,
paleolimbot merged PR #10:
URL: https://github.com/apache/arrow-nanoarrow/pull/10
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
paleolimbot closed issue #4: Implement bitmap helpers
URL: https://github.com/apache/arrow-nanoarrow/issues/4
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe,
paleolimbot opened a new pull request, #14:
URL: https://github.com/apache/arrow-nanoarrow/pull/14
Fixes #5 by implementing an Array whose buffer lifecycle is handled by
`struct ArrowBuffer`.
--
This is an automated message from the Apache Git Service.
To respond to the message, please
lidavidm commented on code in PR #14:
URL: https://github.com/apache/arrow-nanoarrow/pull/14#discussion_r939166568
##
src/nanoarrow/typedefs_inline.h:
##
@@ -165,6 +212,20 @@ struct ArrowBitmap {
int64_t size_bits;
};
+/// \brief A structure used as the private data
paleolimbot commented on PR #10:
URL: https://github.com/apache/arrow-nanoarrow/pull/10#issuecomment-1206514130
I see...I'd been using it to simplify the append process, but the right
thing to do is to properly bitpack-as-you-append (which is now implemented) so
that the `ArrowBufferXXX()`
dependabot[bot] opened a new pull request, #54:
URL: https://github.com/apache/arrow-adbc/pull/54
Bumps [postgresql](https://github.com/pgjdbc/pgjdbc) from 42.4.0 to 42.4.1.
Changelog
Sourced from https://github.com/pgjdbc/pgjdbc/blob/master/CHANGELOG.md;>postgresql's
changelog.
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942434862
##
src/nanoarrow/utils_inline.h:
##
@@ -26,6 +26,114 @@
extern "C" {
#endif
+static inline void ArrowLayoutInit(struct ArrowLayout* layout,
+
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942417008
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942442675
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942439274
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942446454
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942472594
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942437669
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942470780
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942446356
##
src/nanoarrow/array_inline.h:
##
@@ -0,0 +1,146 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942431188
##
src/nanoarrow/typedefs_inline.h:
##
@@ -166,6 +166,24 @@ enum ArrowType {
NANOARROW_TYPE_INTERVAL_MONTH_DAY_NANO
};
+/// \brief Functional types of
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942447804
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
paleolimbot commented on PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#issuecomment-1210685336
Just a note that I'm reworking this interface based on some thoughts after
working with this for a day or so:
- Instead of copying all the buffer/bitmap methods for the
paleolimbot commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942441450
##
src/nanoarrow/nanoarrow.h:
##
@@ -508,9 +512,31 @@ void ArrowArraySetValidityBitmap(struct ArrowArray* array,
struct ArrowBitmap* b
ArrowErrorCode
zeroshade commented on issue #60:
URL: https://github.com/apache/arrow-adbc/issues/60#issuecomment-1211012442
at a minimum it would be good to be able to at least know the *number* of
expected inputs even if the schema isn't knowable.
Maybe having two values? an integer indicating
lidavidm commented on code in PR #16:
URL: https://github.com/apache/arrow-nanoarrow/pull/16#discussion_r942753583
##
src/nanoarrow/array_inline.h:
##
@@ -0,0 +1,246 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.
lidavidm commented on issue #60:
URL: https://github.com/apache/arrow-adbc/issues/60#issuecomment-1211134073
How about this? The parameters are always encoded as a schema, but unknown
types are represented as just NullType. Avoids having lots of optional
things/multiple calls.
codecov-commenter commented on PR #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17#issuecomment-1211146044
#
zeroshade opened a new issue, #60:
URL: https://github.com/apache/arrow-adbc/issues/60
If available, it would be great to be able to retrieve any information about
parameter binding that is available. Some potential information that *might* be
available:
* Number of expected
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1210989539
Also, possibly the driver manager could define execute-with-result-set and
execute-with-rows-affected in terms of the generic execute + generic getters to
retrieve the affected
zeroshade opened a new issue, #59:
URL: https://github.com/apache/arrow-adbc/issues/59
For the common case of executing a single SQL string, let's have a method on
the connection object for executing the query directly without the need for an
intermediate Statement object
--
This is an
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1210994922
CC @pitrou, @hannes, @krlmlr if you have opinions here?
@lwhite1 had the same feedback about executeQuery/execute in Java last
month. So for consistency a query method
zeroshade commented on issue #59:
URL: https://github.com/apache/arrow-adbc/issues/59#issuecomment-1210917005
With this, it might make sense for the `AdbcStatement` object to *only*
represent a prepared statement and place the `Prepare` method on the Connection
rather than on the
lidavidm commented on issue #60:
URL: https://github.com/apache/arrow-adbc/issues/60#issuecomment-1211081338
That makes sense.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
zeroshade opened a new issue, #61:
URL: https://github.com/apache/arrow-adbc/issues/61
Rather than the separate `Execute` / `GetStream` functions, it might be
better to follow something similar to FlightSQL's interface or Go's
`database/sql` API.
Have two functions:
* Execute
lidavidm commented on issue #59:
URL: https://github.com/apache/arrow-adbc/issues/59#issuecomment-1210982209
I think this makes sense to provide as a convenience, but maybe not as the
only method. The separate Statement object still lets us configure any options
in an ABI-compatible way
lidavidm commented on issue #60:
URL: https://github.com/apache/arrow-adbc/issues/60#issuecomment-1210983818
Flight SQL provides this. I think this makes sense, but yeah, something like
`SELECT ?, ?` is going to be dubious. I don't know if there's a great way of
indicating that, though.
paleolimbot opened a new pull request, #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17
It turns out this is really annoying to do otherwise! Declaring a variable
of an appropriate type gets verbose when switching on type, and it sounds like
these functions might be useful for
zeroshade commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1210860838
Two more gaps to add:
* Retrieve the last inserted id for inserts into an auto-increment table
* Retrieve the number of rows affected by the last query (number inserted /
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1210987487
I think it may still have sense to have a generic Execute to ease
compatibility with APIs that do not differentiate between the types of queries
(and note JDBC has all three!), but
lidavidm merged PR #58:
URL: https://github.com/apache/arrow-adbc/pull/58
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
lidavidm commented on code in PR #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17#discussion_r942812080
##
src/nanoarrow/buffer_test.cc:
##
@@ -160,3 +160,31 @@ TEST(BufferTest, BufferTestError) {
ArrowBufferReset();
}
+
+TEST(BufferTest,
zeroshade commented on issue #60:
URL: https://github.com/apache/arrow-adbc/issues/60#issuecomment-1211193521
Seems good to me! :)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
lidavidm commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1211235028
Punting on paramstyle and last inserted ID, but adding row count and current
catalog:
```diff
commit 50b2e40d727c0a51029d7f5506c0696b3a19a3b9
Author: David Li
Date:
lidavidm commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1211236715
Returning strings from a C API is a bit annoying and I'm not sure whether
this is preferable, or if we want to go with an ODBC-style API (pass a
caller-allocated buffer and length
lidavidm commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1211184131
So looking into it
- rowcount is easy to bind, but hard to support (lots of things don't
support it or only support it for inserts) - that's OK. Flight SQL only exposes
it for
paleolimbot commented on PR #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17#issuecomment-1211464135
Ok - this is a first pass at #8 that implements the functions needed to make
"build by buffer" a thing. The `ArrowBufferAppendInt8()` family of functions
helps make code that
paleolimbot commented on code in PR #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17#discussion_r943029428
##
src/nanoarrow/buffer_test.cc:
##
@@ -160,3 +160,31 @@ TEST(BufferTest, BufferTestError) {
ArrowBufferReset();
}
+
+TEST(BufferTest,
paleolimbot merged PR #17:
URL: https://github.com/apache/arrow-nanoarrow/pull/17
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
zeroshade commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1212115628
Is the idea that `RowCount` would do double duty as the number of rows in a
result set OR the number of rows affected by an update/insert?
Given the lack of reliable support,
lidavidm commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1212119635
> Is the idea that `RowCount` would do double duty as the number of rows in
a result set OR the number of rows affected by an update/insert?
Yeah, I don't see a reason to have
zeroshade commented on issue #55:
URL: https://github.com/apache/arrow-adbc/issues/55#issuecomment-1212124555
Sounds good to me
> Also, I would argue these sorts of use cases are mostly out of scope,
though that's mostly my assumption.
I agree, seems fine for that to be out of
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1214417469
I'll chime in from the Go database/sql package:
* A Connection Object is assumed to be Stateful and will not be used
concurrently by multiple Goroutines
* Connections have
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1214424587
Right, and on the other hand, databases like SQLite have no reliable way to
get the info. But APIs like JDBC, Python DBAPI, and Go's database API expose
standard ways to get last
krlmlr commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1214423568
Last inserted IDs (or the results of computed columns in general, for that
matter) can be obtained with the `RETURNING` syntax for most databases, SQL
Server has `OUTPUT` . This seems
krlmlr commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1214402390
Just to be sure we're on the same page:
- a "query" is a single SQL string that can return a result set but doesn't
have to
- a "statement" is the result of preparing an SQL
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1214418597
Thanks, I think that aligns with what I am basically assuming so far: no
concurrent access (though maybe a particular driver can relax this, e.g. Flight
SQL), but also no guarantees
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1214419966
> Just to be sure we're on the same page:
>
> * a "query" is a single SQL string that can return a result set but
doesn't have to
>
> * a "statement" is the
paleolimbot commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1214944399
> no guarantees on thread identity
This is something I've noticed about calls to the `struct ArrowArrayStream`
methods from both Arrow and DuckDB (sometimes they happen on
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215042875
@paleolimbot in that case, wouldn't it be the responsibility of the R/ADBC
bridge to manage the threading? Requiring that all calls to ADBC happen from
the same thread is rather
pitrou commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215072668
> @paleolimbot in that case, wouldn't it be the responsibility of the R/ADBC
bridge to manage the threading?
That sounds reasonable to me (if at all possible). I presume it is
paleolimbot commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215092186
> I presume it is possible to queue R function calls from another thread?
It's an absolute nightmare to do safely without the consumer evaluating the
function call in a
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215092467
So digging around, I would probably vote that for ADBC, we roughly follow
the JDBC/ODBC concurrency/thread safety guarantees.
Here, a statement is just a handle/object to
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215092911
> I just bring it up in case limiting method calls to the thread that
created the connection is not in fact painful, and with the vague feeling that
the sorry sod who has to work
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215148897
It's still worth thinking about, even if the answer is no :)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215277102
> An individual statement can be used multiple times, but result sets cannot
be read concurrently (that is: executing a statement invalidates prior result
sets)
Are we
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215437455
Hmm, but that might impact something like SQLite or DuckDB - I'd have to
test how DuckDB behaves here, though.
In general, my understanding of most APIs is that prepared
lidavidm opened a new pull request, #66:
URL: https://github.com/apache/arrow-adbc/pull/66
Add documentation around thread safety/concurrency/overlapping usage of a
single object.
Sets up Doxygen.
Fixes #64.
--
This is an automated message from the Apache Git Service.
To
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215574770
JDBC/ODBC and Flight SQL/Go are unfortunately at odds here, though. I'm not
sure if there's a great way to express that Flight SQL lets you do this without
penalty without also
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215593890
@lidavidm I was writing that response and hten stepped away for a meeting
and came back to all your comments haha. but TL;DR: looks like under the hood
the database/sql package in
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215591703
> In general, my understanding of most APIs is that prepared statements
aren't really set up to benefit concurrent execution, only repeated execution
from a single logical chain of
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215400906
> I am only talking about a single query, so this is sort of irrelevant to
Go. For Go, I would expect each call to Query to initialize and use a new
AdbcStatement, at which point
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215409446
Flight SQL would be able to do this easily, but libpq can't do this at all,
for instance - that's the balance I'm trying to strike here. (Also even for
Flight SQL, I'm not certain
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215587882
Possibly not, given that most APIs don't (or, they materialize up front).
I'm also still a little skeptical whether Flight SQL truly allows this, or
whether it's just not been
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215594902
I should've read the Go docs a little more closely :sweat_smile: in that
case I'll keep the linked PR as-is, if it looks reasonable. Thanks for the help!
--
This is an automated
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215589955
Hmm. Go's docs state this for Stmt:
> When the Stmt needs to execute on a new underlying connection, it will
prepare itself on the new connection automatically.
paleolimbot commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215146633
No worries then - ignore me!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the
lidavidm commented on PR #66:
URL: https://github.com/apache/arrow-adbc/pull/66#issuecomment-1215532046
I guess DuckDB materializes the results fully when you query so that's not a
worry (we could support Go's behavior)
--
This is an automated message from the Apache Git Service.
To
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215562560
It will still create difficulties for drivers that wish to wrap JDBC/ODBC,
which explicitly disallow this, though. Those will have to materialize all
results in memory.
--
This
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215470298
Draft of some clarifying text in #66
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to
lidavidm commented on PR #66:
URL: https://github.com/apache/arrow-adbc/pull/66#issuecomment-1215469871
TODO here: reconcile with Go's prepared statement behavior in #64 and see
what DuckDB does in the described scenario
--
This is an automated message from the Apache Git Service.
To
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215342377
> Are we referring to result sets from _different queries_ or only putting a
limitation on result sets from a single prepared statement?
So to be clear, "statement" here
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215542812
Long story short @zeroshade I'll tweak #66 to clarify that result sets are
not invalidated except by closing the statement, but to caution that some
drivers may have to materialize
pitrou commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215567684
Having to materialize all results instead of streaming them sounds a rather
major pitfall.
--
This is an automated message from the Apache Git Service.
To respond to the message,
pitrou commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215580804
> I'm not sure if there's a great way to express that Flight SQL lets you do
this [...]
Is it important to express it?
--
This is an automated message from the Apache Git
lidavidm commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1215630552
Ok, so how does this sound (here, I'm ignoring #59 provide a "just query"
method):
Remove `AdbcStatementGetStream`
Change `AdbcStatementExecute` to
wesm opened a new issue, #20:
URL: https://github.com/apache/arrow-nanoarrow/issues/20
In the spirit of making vendoring simpler for users, I'm wondering what we
can do to reduce the total number of files in the project. In very large
projects that are intending to be vendored (like DuckDB
lidavidm opened a new issue, #21:
URL: https://github.com/apache/arrow-nanoarrow/issues/21
Just to prove our claim of being easily embeddable, and to make sure that
things work if multiple versions of nanoarrow end up in a binary (e.g. through
two different dependencies)
--
This is an
lidavidm commented on code in PR #65:
URL: https://github.com/apache/arrow-adbc/pull/65#discussion_r946050839
##
c/drivers/postgres/statement.cc:
##
@@ -0,0 +1,283 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements. See
zeroshade commented on issue #61:
URL: https://github.com/apache/arrow-adbc/issues/61#issuecomment-1215736580
That seems reasonable to me @lidavidm
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go
lidavidm commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215412769
I suppose we could keep the ArrowArrayStream as independent, and that would
require `Execute` to block and accumulate all results in the case of libpq
(which is also probably
zeroshade commented on issue #64:
URL: https://github.com/apache/arrow-adbc/issues/64#issuecomment-1215600926
@lidavidm Don't beat yourself up, it wasn't in the docs haha. I had to read
the actual source code to figure that out :smile:
--
This is an automated message from the Apache Git
lidavidm commented on issue #21:
URL: https://github.com/apache/arrow-nanoarrow/issues/21#issuecomment-1216068768
jemalloc lets you do it as well (it's used in Arrow C++), though I'm not
sure how it works there. (In that case, symbols are renamed, but still
accessible.)
I think this
wesm commented on issue #20:
URL: https://github.com/apache/arrow-nanoarrow/issues/20#issuecomment-1216074095
That seems odd that VSCode would struggle with large files (should this be
reported as a bug?). It does not seem like a reason to compromise the design of
the library.
--
This
paleolimbot commented on issue #20:
URL: https://github.com/apache/arrow-nanoarrow/issues/20#issuecomment-1216058824
I think we could get it down to something like that...off the top of my head
I can see 2 .h and 3 .c making sense (nanoarrow.h, nanarrow_inline.h, array.c,
schema.c,
paleolimbot commented on issue #21:
URL: https://github.com/apache/arrow-nanoarrow/issues/21#issuecomment-1216064759
I was thinking that an `examples/` directory should exist...this should
definitely be one of them!
With respect to symbol renaming, I've only seen this in H3:
lidavidm commented on issue #18:
URL: https://github.com/apache/arrow-nanoarrow/issues/18#issuecomment-1212365892
Hmm, wouldn't you just validate the lengths once in `ArrrayViewInit`, and
then the getters could just validate the given index against the logical length
already in ArrowArray?
bdklahn opened a new issue, #330:
URL: https://github.com/apache/arrow-julia/issues/330
I'm so glad someone implemented Arrow for Julia. Thanks!
And I think the intro to the User Manual is the clearest I've come across to
help understand the what and why of Arrow.
It looks
paleolimbot commented on issue #18:
URL: https://github.com/apache/arrow-nanoarrow/issues/18#issuecomment-1212370668
I was separating the `ArrowLayout` bit into its own PR today anyway and will
play with it a bit...maybe the `ArrowBufferView is part of the `ArrayView` so
that it's only
lidavidm opened a new issue, #18:
URL: https://github.com/apache/arrow-nanoarrow/issues/18
This isn't strictly necessary - but I wonder what you'd think of having
inline getters, things like
```c
int ArrowArrayIsSet(struct ArrowArray* array, int64_t i) {
// TODO: unions
paleolimbot commented on issue #18:
URL: https://github.com/apache/arrow-nanoarrow/issues/18#issuecomment-1212348074
I think that makes sense! Although it may need some extra information
(storage type + layout) to be generic. Maybe:
```c
struct ArrowArrayView(struct ArrowArray*
paleolimbot opened a new pull request, #19:
URL: https://github.com/apache/arrow-nanoarrow/pull/19
Implements a few types whose goal is to provide access to buffers. This will
require a `switch(storage_type)` by the consumer and requires some knowledge of
the spec (getters to parallel the
lidavidm commented on code in PR #19:
URL: https://github.com/apache/arrow-nanoarrow/pull/19#discussion_r943929415
##
src/nanoarrow/utils_inline.h:
##
@@ -26,6 +26,115 @@
extern "C" {
#endif
+static inline void ArrowLayoutInit(struct ArrowLayout* layout,
Review Comment:
601 - 700 of 1547 matches
Mail list logo