func prepareTestBackend(ctx context.Context, instanceType string) (ObjectLayer, []string, error) {
	switch instanceType {
	// Total number of disks for Erasure sets backend is set to 32.
	case ErasureSetsTestStr:
		return prepareErasureSets32(ctx)
	// Total number of disks for Erasure backend is set to 16.
	case ErasureTestStr:
		return prepareErasure16(ctx)
	default:
		// return FS backend by default.

   * queue.
   */
  public static <E> EvictingQueue<E> create(int maxSize) {
    return new EvictingQueue<>(maxSize);
  }

  /**
   * Returns the number of additional elements that this queue can accept without evicting; zero if
   * the queue is currently full.
   *
   * @since 16.0
   */
  public int remainingCapacity() {
    return maxSize - size();
  }

     * down. twiceSignifFloor will contain the top SIGNIFICAND_BITS + 2 bits, and signifFloor the
     * top SIGNIFICAND_BITS + 1.
     *
     * It helps to consider the real number signif = absX * 2^(SIGNIFICAND_BITS - exponent).
     */
    int shift = exponent - SIGNIFICAND_BITS - 1;
    long twiceSignifFloor = absX.shiftRight(shift).longValue();
    long signifFloor = twiceSignifFloor >> 1;

# ADR-0004 - Use a platform-oriented architecture for Gradle

## Date

2024-02-07

## Context

The Gradle code base is essentially a large monolith, without strong internal boundaries.
This has a number of negative effects on productivity, including:

- Unclear ownership of code.
- Difficult to focus on one particular area.
- Unintended coupling between areas of the code, including tests.

## Decision

Prefer either:
  * Provide only non-nullable version and move the `null` check to the call site (if the number of nullable callsites is low).
  * Provide two overloads with a different names (as Java doesn't support overloading on nullability). See 

      out.write(bytes);
    }
  }

  /**
   * Writes all the bytes from the given {@code InputStream} to this sink. Does not close {@code
   * input}.
   *
   * @return the number of bytes written
   * @throws IOException if an I/O occurs while reading from {@code input} or writing to this sink
   */
  @CanIgnoreReturnValue
  public long writeFrom(InputStream input) throws IOException {

      out.write(bytes);
    }
  }

  /**
   * Writes all the bytes from the given {@code InputStream} to this sink. Does not close {@code
   * input}.
   *
   * @return the number of bytes written
   * @throws IOException if an I/O occurs while reading from {@code input} or writing to this sink
   */
  @CanIgnoreReturnValue
  public long writeFrom(InputStream input) throws IOException {

| xlHeaderVersion | msgp uint                   | header version identifier
| xlMetaVersion   | msgp uint                   | metadata version identifier
| versions        | msgp int                    | Number of versions following
| header_1        | msgp bin array              | Header of version 1
| metadata_1      | msgp bin array              | Metadata of version 1

		Results: make([]int, len(fi.Parts)),
	}

	erasure := fi.Erasure
	for i, part := range fi.Parts {
		checksumInfo := erasure.GetChecksumInfo(part.Number)
		partPath := pathJoin(volumeDir, path, fi.DataDir, fmt.Sprintf("part.%d", part.Number))
		err := s.bitrotVerify(ctx, partPath,
			erasure.ShardFileSize(part.Size),
			checksumInfo.Algorithm,
			checksumInfo.Hash, erasure.ShardSize())

* Adding labels to PRs.
* Editing PR titles.
* Adding commits on top of PRs to tweak them.
* Mark answers in GitHub Discussions questions, etc.
* Merge some specific types of PRs.

You can see the current team members in [FastAPI People - Team](./fastapi-people.md#team){.internal-link target=_blank}.

Joining the team is by invitation only, and I could update or remove permissions, instructions, or membership.

## FastAPI Experts