Add `CreateObject` function which slices data stream into NeoFS objects
and saves the produced objects in the NeoFS network using `Client`.
Closes#342.
Signed-off-by: Leonard Lyubich <leonard@morphbits.io>
In previous implementation `ObjectWrite.Close` called `close` method
which in turn called `result` callback. Thus failed statuses could lead
to false-positive result processing.
Replace calling `close` method with direct `closer` method's call in
`ObjectWrite.Close`.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous implementation `client` package provided access to nested
response fields as pointers to them. This caused clients to handle nil
cases even when the field presence in the response is required.
Avoid returning pointers to required fields in response getters. This
also reduces reference counter load and allows fields to be decoded
directly without additional assignment.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
After 85e3c7b087 `processResponse` method
sets `err` field or sets status in `statusRes` field of `contextCall`.
In previous implementation `ObjectWriter.Close` method returned
`ctxCall.err` on `false` return of `processResponse` method. This could
cause NPE-panic if status failure resolving was disabled.
Make `ObjectWriter.Close` to return internal `err` field only if it is
set, otherwise return status response.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
Object service of NeoFS API contains one client-side stream method:
object.Put. In client-side streams, server can return an error after
processing stream message. In this case write method returns `io.EOF`
and actual error reason is encoded in response status, which is
obtained after `Close()`. Client library should process such case.
Signed-off-by: Alex Vanin <alexey@nspcc.ru>
Get rid of `Option` pattern. Define `Init`, `Dial` and `Close` methods
for the corresponding stages of use.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
Extend docs with supported status returns. Add several helper functions
which allow to check the particular status.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous implementation payload chunks were split into pieces with
512B length. This led to sending a large number of messages with a large
amount of payload.
Increase per-message payload chunk limit to 3MB.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
