buffer: let WriteFloatGeneric silently drop values

[pmq20]

Fixes#3766

Quoted from buffer.markdown,

Set noAssert to true to skip validation of value and offset. This means
that value may be too large for the specific function and offset may be
beyond the end of the buffer leading to the values being silently dropped.

Therefore I let node::Buffer::WriteFloatGeneric silently drop values instead of crashing.

[thefourtheye]

cc @trevnorris

[trevnorris]

Just curious, why initialize memcpy_num to sizeof(T)?

[pmq20]

Just a coding style thing, it is the same as,

if (offset + sizeof(T) > ts_obj_length)
memcpy_num = ts_obj_length - offset;
else
memcpy_num = sizeof(T);

[trevnorris]

ah I see. was focused on that memcpy_num = sizeof(T) was set then immediately after there's a check sizeof(T) in the conditional. nm it then. :)

[trevnorris]

@pmq20 hm. possibly the index returned should be the end of the buffer and not just index + 8. don't really care though. Everything LGTM.

CI is down ATM. will queue this up when it becomes available. Thanks much for the fix!

[jasnell]

@trevnorris ... would this be a bug fix or a behavior change? Think this would be appropriate for LTS?

[trevnorris]

@jasnell Since the alternative is to flat out abort, I'd consider this a bug fix.

[pmq20]

Thank you all for reviewing. Please let me know if I could do more.

[thefourtheye]

Comment says BE, but the code uses LE. May be add test for that also.

[trevnorris]

@pmq20 If you wouldn't mind adding similar tests to cover all the write{Double,Float}{LE,BE} variants (ping me when you do) I'll put this through CI and land it.

[pmq20]

@trevnorris done

[trevnorris]

CI: https://ci.nodejs.org/job/node-test-pull-request/753/

[pmq20]

@trevnorris The failures are on ARM:

 ...
not ok 132 test-crypto-dh.js
# TIMEOUT
---
duration_ms: 120.25

not ok 129 test-crypto-binary-default.js
# TIMEOUT
---
duration_ms: 120.101

They seem unrelated to this PR.

[trevnorris]

Landed in 0ed3a7c. Thanks much!