doc: fix encoding string in buffer example

[MapleGu]

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• documentation is changed or added
• commit message follows commit guidelines

Affected core subsystem(s)

[jseijas]

Hi! In the write method of Buffer there is a switch for the encoding, where the different encoding are taked into account with dash and without dash. So bot utf8 and utf-8 works, ucs2 and ucs-2, utf16le and utf-16le... I think that the best solution is not to modify the documentation, but adding the latin-1 option in the "case 7" of the switch:

 case 7:
      if (encoding === 'utf16le' || encoding.toLowerCase() === 'utf16le')
        return this.ucs2Write(string, offset, length);
      /* begin modification */
      if (encoding === 'latin-1')
        return this.latin1Write(string, offset, length);
      /* end modification */
      break;
    case 8:
      if (encoding === 'utf-16le' || encoding.toLowerCase() === 'utf-16le')
        return this.ucs2Write(string, offset, length);
      break;
    case 6:
      if (encoding === 'latin1' || encoding === 'binary')
        return this.latin1Write(string, offset, length);

[vsemozhetbyt]

cc @nodejs/buffer

[addaleax]

Thank you! It would be great if you could reword the commit message so that it starts with doc: and describes what the patch itself does, e.g. doc: fix encoding string in buffer example. If you don’t feel comfortable doing that we can also do that for you when landing the commit. :)

[addaleax]

@jseijas I would be careful about adding another encoding alias; the ones we have right now are mostly there for historic reasons, and having to support more encoding strings comes with a certain performance impact.

That we have introduced latin1 as an alias of binary in the first place shows that there are obviously exceptions, of course. ;)

(Also, there’s quite a few more places in the codebase where such an alias would have to be supported.)

[jasnell]

+1 to @addaleax here... adding a new alias carries a definite non-zero cost in terms of complexity and performance for very little gain.

[jasnell]

The doc change is trivial enough that this shouldn't need to wait 48 hours to land

[TimothyGu]

Landed in 3bf358d.