Alignas broken when used with constexpr array data member for structure
Hi there,
In GCC 4.8.4 I have something like the following:
constexpr int x = 5;
constexpr int y = 4;
struct alignas(y) my_data_block {
char data[x];
};
And it causes some weird errors to the tune of "size of array ‘data’ is
not an integral constant-expression" in the presence of the alignas
This is a pretty nasty bug and means it's not implemented as I thought.
I don't know the front-ends (but I do actually know GIMPLE-low and below
quite well, love the pattern matching) and I'd like to dig more, it's
almost certainly fixed - this is just for personal curiosity.
Where would I look? A 1 line reply with a directory would be a great
start; even if it's just a guess.
I did ask in #gcc on freenode - it didn't go so well, sorry to ping you
all for this.
Alec
Re: Alignas broken when used with constexpr array data member for structure
On 9 January 2018 at 13:15, Alec Teal wrote:
> Hi there,
>
> In GCC 4.8.4 I have something like the following:
>
> constexpr int x = 5;
>
> constexpr int y = 4;
>
> struct alignas(y) my_data_block {
>
>char data[x];
>
> };
>
>
> And it causes some weird errors to the tune of "size of array ‘data’ is not
> an integral constant-expression" in the presence of the alignas
>
> This is a pretty nasty bug and means it's not implemented as I thought. I
> don't know the front-ends (but I do actually know GIMPLE-low and below quite
> well, love the pattern matching) and I'd like to dig more, it's almost
> certainly fixed - this is just for personal curiosity.
>
> Where would I look? A 1 line reply with a directory would be a great start;
> even if it's just a guess.
I told you on IRC, C++ support is implemented in gcc/cp
Why do you keep demanding help and then ignore the answer?
> I did ask in #gcc on freenode - it didn't go so well, sorry to ping you all
> for this.
This belongs on gcc-help not this list.
Re: Implementing p0515 - spaceship operator
On 01/08/2018 02:07 PM, Tim van Deurzen wrote: Hi, I've been spending some time the past few weeks implementing p0515r2, i.e. the proposal for consistent comparisons for C++ (aka the spaceship operator). I've received some very valuable help on the IRC channel, but I'm still a little bit stuck. Note, I'm completely new to the GCC codebase and am very much still getting oriented. Following advice from some of the people in IRC channel I implemented parsing the new token in libcpp and I seem to be successfully parsing it and creating a new AST node. As this feature is not in the C++ standard yet, I wanted to add a command line flag to enable usage of the new operator and ignoring it otherwise. I managed to get cc1plus to accept the parameter, but it seems that when I invoke my own g++ binary with that parameter, it complains about unknown parameters. As Jonathan mentioned, you may not need a new option for this. But it's useful to know how to add one in any case. Reading the manual is a good idea but the easiest way to actually add one by far is to look at a simple change that adds in git or svn log one and follow its example. This command will give you such a list: git log -p --grep="New option" Then just look for an option that affects the front-end. A warning options might be a good example. Martin I'm perfectly happy to dig further on my own, but I get the feeling I'm missing some documentation / resource somewhere that might help me out. Is there some documentation about adding and passing around parameters that will be used both in libcpp and the C++ front-end? What would be the best place to look to learn more about how part of GCC this is structured? I want to make sure I go about this correctly. If this is the wrong place to ask for help, please redirect me, so that I don't unnecessarily spam the wrong mailing list :-). Kind regards, Tim.
Re: Implementing p0515 - spaceship operator
On 01/08/2018 05:07 PM, Jason Merrill wrote: On Mon, Jan 8, 2018 at 4:07 PM, Tim van Deurzen wrote: There's a gccint.info documentation file, the Options node seems like what you're looking for. You will want to add a new option to c-family/c.opt. diffing between branches/c++-modules and trunk may be educational, as that has to do similar things to enable the new keywords. nathan -- Nathan Sidwell
Re: Implementing p0515 - spaceship operator
On 01/08/2018 11:28 PM, Jason Merrill wrote: On Mon, Jan 8, 2018 at 5:13 PM, Jonathan Wakely wrote: On 8 January 2018 at 22:07, Jason Merrill wrote: On Mon, Jan 8, 2018 at 4:07 PM, Tim van Deurzen wrote: I've been spending some time the past few weeks implementing p0515r2, i.e. the proposal for consistent comparisons for C++ (aka the spaceship operator). Great! I've received some very valuable help on the IRC channel, but I'm still a little bit stuck. Note, I'm completely new to the GCC codebase and am very much still getting oriented. Following advice from some of the people in IRC channel I implemented parsing the new token in libcpp and I seem to be successfully parsing it and creating a new AST node. As this feature is not in the C++ standard yet, I wanted to add a command line flag to enable usage of the new operator and ignoring it otherwise. I managed to get cc1plus to accept the parameter, but it seems that when I invoke my own g++ binary with that parameter, it complains about unknown parameters. I'm perfectly happy to dig further on my own, but I get the feeling I'm missing some documentation / resource somewhere that might help me out. Is there some documentation about adding and passing around parameters that will be used both in libcpp and the C++ front-end? What would be the best place to look to learn more about how part of GCC this is structured? I want to make sure I go about this correctly. There's a gccint.info documentation file, the Options node seems like what you're looking for. You will want to add a new option to c-family/c.opt. operator<=> is now in the working paper though, so does it need its own option? IMHO it should just be enabled for -std=c++2a and -std=gnu++2a, and not otherwise. Ah, good point. Jason That makes sense, I'll just use the c++2a and gnu++2a flags. Though I'll still take some time to figure out how options work, so that I can add any toggles that (e.g. for warnings) that might be specific to operator<=>. Just to confirm with you, it does make sense to conditionally parse the token for operator<=> in libcpp (i.e. only when the cxx standard being used is >=2a)? I'm just wondering if this does not accidentally affect other front-ends using libcpp? Tim.
