Re: Can't interrupt tight loops in Lua 5.4 anymore?

[Viacheslav Usov <via.usov@...>]

On Sat, May 23, 2020 at 7:48 PM Roberto Ierusalimschy
<roberto@inf.puc-rio.br> wrote:

> Looking at that commit, we can see that we have had more than two
> years to discuss that design. It was available for comments in the
> work versions, the alpha version, and the beta version of Lua 5.4.

This is certainly true. However, this particular change was not one
listed in the "main changes" of the nearest work releases, and was
not, as far as I can tell, highlighted in any other way. And back then
the discussions focused on more prominent things, such as nils in
tables, the recursive vs "stackless" interpreter, etc. I am not trying
to imply that anything was wrong with that, just that it is naturally
difficult to understand implications of every change in the Lua source
code, especially of those as technical as the one in question.

> Now that we are releasing Lua 5.4, we are focused on fixing bugs, not in
> improving its design.

I cannot call the potential problem described in the first message in
this thread a bug because, as far as I know, no one has yet reported
this as something they really encountered. Yet, this kind of problems
is much more likely to surface in production rather than in any
preliminary testing, so I think that fixing this potential problem now
would be good.

Whether this should be fixed in a manner proposed in Joseph's patches,
or in some other way, well, I obviously have some opinions about that,
but this is far less important than having this fixed, or, rather,
prevented. That said, the elimination of gyrations related to
coroutines when one has to interrupt a running script that may have
used them, as afforded by the patches, would be most certainly very
welcome.

The request for clarification in my previous email should not be
interpreted to imply that I request any design changes for 5.4.
Obviously, if I am given a clarification, I intend to think it through
and possibly suggest something, but that would be tertiary in
importance.

Cheers,
V.