Re: [PATCH 0/2] add unique_ptr class

[Pedro Alves <palves@redhat.com>]

On 09/04/2017 11:31 AM, Richard Biener wrote:
> On Fri, Aug 11, 2017 at 10:43 PM, Jonathan Wakely <jwakely@redhat.com> wrote:
>> On 05/08/17 20:05 +0100, Pedro Alves wrote:
>>>
>>> On 08/04/2017 07:52 PM, Jonathan Wakely wrote:
>>>>
>>>> On 31/07/17 19:46 -0400, tbsaunde+gcc@tbsaunde.org wrote:
>>>>>
>>>>> I've been saying I'd do this for a long time, but I'm finally getting to
>>>>> importing the C++98 compatable unique_ptr class Pedro wrote for gdb a
>>>>> while
>>>>> back.
>>>
>>>
>>> Thanks a lot for doing this!
>>>
>>>> I believe the gtl namespace also comes from Pedro, but GNU template
>>>> library seems as reasonable as any other name I can come up with.
>>>
>>>
>>> Yes, I had suggested it here:
>>>
>>>  https://sourceware.org/ml/gdb-patches/2017-02/msg00197.html
>>>
>>>>
>>>> If it's inspired by "STL" then can we call it something else?
>>>>
>>>> std::unique_ptr is not part of the STL, because the STL is a library
>>>> of containers and algorithms from the 1990s. std::unique_ptr is
>>>> something much newer that doesn't originate in the STL.
>>>>
>>>> STL != C++ Standard Library
>>>
>>>
>>> That I knew, but gtl was not really a reference to the
>>> C++ Standard Library, so I don't see the problem.  It was supposed to
>>> be the name of a library which would contain other C++ utilities
>>> that would be shared by different GNU toolchain components.
>>> Some of those bits would be inspired by, or be straight backports of
>>> more-recent standards, but it'd be more than that.
>>>
>>> An option would be to keep "gtl" as acronym, but expand it
>>> to "GNU Toolchain Library" instead.
>>
>>
>> OK, that raises my hackles less. What bothered me was an apparent
>> analogy to "STL" when that isn't even the right name for the library
>> that provides the original unique_ptr.
>>
>> And "template library" assumes we'd never add non-templates to it,
>> which is unlikely (you already mentioned offset_type, which isn't a
>> template).
>>
>> It seems odd to make up a completely new acronym for this though,
>> rather than naming it after something that exists already in the
>> toolchain codebases.
>>
>>
>>> For example, gdb has been growing C++ utilities under gdb/common/
>>> that might be handy for gcc and other projects too, for example:
>>>
>>> - enum_flags (type-safe enum bit flags)
>>> - function_view (non-owning reference to callables)
>>> - offset_type (type safe / distinct integer types to represent offsets
>>>                into anything addressable)
>>> - optional (C++11 backport of libstdc++'s std::optional)
>>> - refcounted_object.h (intrusively-refcounted types)
>>> - scoped_restore (RAII save/restore of globals)
>>> - an allocator that default-constructs using default-initialization
>>>   instead of value-initialization.
>>>
>>> and more.
>>>
>>> GCC OTOH has code that might be handy for GDB as well, like for
>>> example the open addressing hash tables (hash_map/hash_table/hash_set
>>> etc.).
>>>
>>> Maybe Gold could make use of some of this code too.  There
>>> are some bits in Gold that might be useful for (at least) GDB
>>> too.  For example, its Stringpool class.
>>>
>>> I think there's a lot of scope for sharing more code between the
>>> different components of the GNU toolchain, even beyond general
>>> random utilities and data structures, and I'd love to see us
>>> move more in that direction.
>>>
>>>> If we want a namespace for GNU utilities, what's wrong with "gnu"?
>>>
>>>
>>> That'd be an "obvious" choice, and I'm not terribly against it,
>>> though I wonder whether it'd be taking over a name that has a wider
>>> scope than intended?  I.e., GNU is a larger set of projects than the
>>> GNU toolchain.  For example, there's Gnulib, which already compiles
>>> as libgnu.a / -lgnu, which might be confusing.  GCC doesn't currently
>>> use Gnulib, but GDB does, and, there was work going on a while ago to
>>> make GCC use gnulib as well.
>>
>>
>> Good point. "gnutools" might be more accurate, but people might object
>> to adding 10 extra characters for "gnutools::".
>>
>> Naming is important, especially for a whole namespace (not just a
>> single type) so I do think it's worth spending time getting it right.
>>
>> But I could live with gtl as long as nobody ever says "GTL is the GNU
>> STL" or mentions "gtl" and STL in the same breath :-)
> 
> If it should be short use g::.  We can also use gnu:: I guess and I
> agree gnutools:: is a little long (similar to libiberty::).  Maybe
> gt:: as a short-hand for gnutools.

Exactly 3 letters has the nice property of making s/gtl::foo/std::foo/ super
trivial down the road; you don't have to care about reindenting stuff
[1].  Also makes gdb->gtl and gcc->gtl renamings trivial in the same way.
Really a minor thing in the grand scheme of things, but just a FYI that that
factored in a bit in the original motivation for the "gtl" naming back when
I proposed it on the gdb list.

[1] - [PATCH] gdb::{unique_ptr,move} -> std::{unique_ptr,move}:
https://sourceware.org/ml/gdb-patches/2016-11/msg00200.html

Thanks,
Pedro Alves