> ..."string_view", which I don't think the javascript library supports...

I think that's right. The Arrow documentation contains an
Implementation Status page [1] which you can reference to see what
types [2] and other features each implementation supports.

[1] https://arrow.apache.org/docs/status.html
[2] https://arrow.apache.org/docs/status.html#data-types


On Tue, Aug 20, 2024 at 1:52 AM Simon Knight <[email protected]> wrote:
>
> Hello,
> thanks for confirming that. I dug into that, and think the issue is that I 
> was writing from polars, which defaults to writing strings using 
> "string_view", which I don't think the javascript library supports.
> If I write out using pandas, it all seems to work correctly.
>
> Thanks
>
> On Tue, 20 Aug 2024 at 17:16, Bryce Mecum <[email protected]> wrote:
>>
>> Your Arrow JS code looks fine, you may be running into browser
>> security (i.e., CORS). How is the file being hosted? What do you get
>> when you take Arrow JS out of the equation and just print the fetch
>> response (like with response.text())?
>>
>>
>> On Mon, Aug 19, 2024 at 11:48 PM Simon Knight <[email protected]> wrote:
>> >
>> > Hi,
>> > I am using the example from https://arrow.apache.org/docs/js/ to load a 
>> > file from a URL in Chrome
>> >
>> > const table = await tableFromIPC(fetch("/simple.arrow"));
>> >
>> > but get
>> > uncaught (in promise) Error: Unrecognized type: "undefined" (24)
>> >     at tk (apache-arrow:287:260)
>> >     at U.fk [as decode] (apache-arrow:283:430)
>> >     at S.hk [as decode] (apache-arrow:281:236)
>> >     at Fh.decode (apache-arrow:197:110)
>> >     at Zk.Wa (apache-arrow:313:501)
>> >     at Zk.open (apache-arrow:312:317)
>> >     at Zk.pa (apache-arrow:314:6)
>> >     at Zk.next (apache-arrow:308:122)
>> >     at Rk.readAll (apache-arrow:303:117)
>> >     at Dl (apache-arrow:349:635)
>> >
>> > If I download the file manually, and open using pyarrow it loads correctly.
>> >
>> > with pa.OSFile('Downloads/small.arrow', 'rb') as source:
>> >    loaded_array = pa.ipc.open_file(source).read_all()
>> >
>> > The documentation is unclear on how to debug this, any help would be 
>> > appreciated.
>> > Thanks

Reply via email to