[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RFC: c++: provide control over the stack.hh file name

From: Hans Åberg
Subject: Re: RFC: c++: provide control over the stack.hh file name
Date: Fri, 28 Sep 2018 14:36:58 +0200

> On 28 Sep 2018, at 13:55, Frank Heckenbach <address@hidden> wrote:
> Hans Åberg wrote:
>>> On 28 Sep 2018, at 12:11, Akim Demaille <address@hidden> wrote:
>>> I know, that's why we chose deques initially.  
>>> But in reality, given
>>> that we have to copy (or move) to/from the stack to the actions, I'm
>>> not sure this constraint really makes sense.
>> You can skip it, as for me.
> Wow, just wow!
> Back then, you kept insisting on deque through many mails (while
> misunderstanding the difference between (a) moves required in Bison
> generated code, (b) moves done internally by a container class and
> (c) moves in user actions; it can all be read in the archives), you
> kept praising the advantages of deque
> (
> and when I finally gave in and implemented it, you were like "Good."
> (,
> and now just: "You can skip it, as for me." Thanks!

Deque looked good because C++ uses that for stack, but Bison parsers look below 
top so less efficient, though applications tend to spend most time in the 
actions and the lexer and not the parser proper. And now you can move, so 
pre-C++11 and classes that can both not copy and neither move seems rather 
limited in use.

> Remind me never to follow any suggestion by you (or enter into a
> discussion with you) again. Could have saved me a lot of time!

Do as you wish. Now, there is an active maintainer you might tune into, which 
was not the case back then.

> FTR, Akim, I don't care much for deque either; I use it in some
> other places in my code where it makes sense, but I don't see a big
> point using it in Bison, like I said back then, so you can skip it
> for my sake, too.

Perhaps an agreement then!?

reply via email to

[Prev in Thread] Current Thread [Next in Thread]