Re: Enabling --enable-check-lisp-object-type by default on x86 and AMD64

[Paul Eggert]

Eli Zaretskii wrote:

> Maybe we should poll this list about this

In effect we've already probed about it in this thread. Others are welcome to 
chime in.

> even the latest GDB can be built without Python. It's a configure-time option.

The configure-time option defaults to "yes" if Python is installed, which it 
typically is nowadays. It's been "yes" in the major GNU/Linux distributions for 
years, and even MinGW now ships a Python-enabled GDB. So this should not be a 
major problem for developers in practice.

> I'm not sure I like this XIL(something) feature.  For starters, it
> deviates from the format I'm used to, and takes precious screen
> estate.

One gets used to it, and it has the advantage of easily distinguishing 
Lisp_Object from integer values when debugging, which is worth some screen real 
estate.

> It also might confuse people who know GDB but won't
> immediately grasp what "XIL" is.

The confusion won't last long, and any confusion is outweighed by the clarity 
caused by GDB's now distinguishing Lisp_Object values from C integers. I don't 
see this as turning into a real problem, but if I'm wrong I suppose we can 
rename XIL to something more mnemonic.

>   (gdb) p Qnil
>   $2 = 0
>   (gdb) p ignore
>   $3 = XIL(0)

Ah, I didn't think about pretty-printing constants. I fixed that by installing 
the first attached patch. I also tried out a somewhat-older GDB implementation 
(Ubuntu LTS) and fixed some other glitches in the second attached patch. 
However, we shouldn't need to spend a lot of time worrying about glitches with 
ancient debuggers, as we can expect developers to be reasonably up-to-date.

> something whose value is zero is numerically equal to something
> whose value is XIL(0).

Sure, because XIL(0) == 0 when --enable-lisp-object-types=no. Pretty-printing 
doesn't change that.

> As for "some day the pretty-printing could be fancier": what exactly
> did you want to implement?

I wasn't planning anything exactly. The general principle is that the GDB 
pretty-printer should output a string that you can paste back into GDB. (In 
contrast, the long-established "pp" command should output a string that you can 
paste into Elisp.) GDB could pretty-print the Lisp integer 27 as 
"make_number(27)", for example.

0001-Pretty-print-const-Lisp_Objects-in-.gdbinit.patch
Description: Text Data

0002-Port-.gdbinit-to-GDB-7.11.1-Python-2.7.12.patch
Description: Text Data