GNAT on GCC 10.x has build problems

GNAT on GCC 10.x has build problems

[Dave Blanchard]

Hello all, while bootstrapping GNAT onto my cross compiled system with GCC 10.x I found that the make script leaves something to be desired. 

First off it doesn't add the host prefix to the cross compiler binaries; it calls gnatmake, gnatlink, gnatbind, gnatls, and gcc without the x86_64-linux-gnu- prefix, requiring an ugly hack to symlink to those tools to complete the build. Also at the end of the build there is an error when it tries to install gnatdll, which isn't built, doesn't exist, and is for Windows only. 

Does anyone know if these problems are fixed in later GCC versions? And do these people even test their obviously broken crap before releasing it on the world?

I also note that the AdaCore team seem to be doing everything in their power to railroad people into giving them money for their proprietary compiler. First they disabled C and C++ languages in their 2018-up Community Edition binaries, which makes bootstrapping a full GCC impossible with these tools. I guess that didn't do the trick as people just used older CE releases to bootstrap with instead, so they discontinued CE entirely and removed links to the CE download page from their site, making it hard to find unless one knows what to search for. 

I've also been told that there is some kind of special licensing clause for the GNAT project which requires all code built by their GPL compiler to be GPL3 licensed, which is a laugh as I'm never doing that. Not sure if that's actually true or not. 

Anyhow, it's surprising (or should be surprising) to see such shoddy workmanship from an anti-freedom commercial organization joined to the hip with GCC.

Dave

Re: GNAT on GCC 10.x has build problems

[Jonathan Wakely]

[-- Attachment #1: Type: text/plain, Size: 1908 bytes --]

On Sun, 3 Dec 2023, 18:19 Dave Blanchard, <dave@killthe.net> wrote:

> Hello all, while bootstrapping GNAT onto my cross compiled system with GCC
> 10.x I found that the make script leaves something to be desired.
>
> First off it doesn't add the host prefix to the cross compiler binaries;
> it calls gnatmake, gnatlink, gnatbind, gnatls, and gcc without the
> x86_64-linux-gnu- prefix, requiring an ugly hack to symlink to those tools
> to complete the build. Also at the end of the build there is an error when
> it tries to install gnatdll, which isn't built, doesn't exist, and is for
> Windows only.
>
> Does anyone know if these problems are fixed in later GCC versions? And do
> these people even test their obviously broken crap before releasing it on
> the world?
>

You know it's possible to ask for help without acting this way?

However, please use the gcc-help list to do so, not this one.


> I also note that the AdaCore team seem to be doing everything in their
> power to railroad people into giving them money for their proprietary
> compiler. First they disabled C and C++ languages in their 2018-up
> Community Edition binaries, which makes bootstrapping a full GCC impossible
> with these tools. I guess that didn't do the trick as people just used
> older CE releases to bootstrap with instead, so they discontinued CE
> entirely and removed links to the CE download page from their site, making
> it hard to find unless one knows what to search for.
>
> I've also been told that there is some kind of special licensing clause
> for the GNAT project which requires all code built by their GPL compiler to
> be GPL3 licensed, which is a laugh as I'm never doing that. Not sure if
> that's actually true or not.
>
> Anyhow, it's surprising (or should be surprising) to see such shoddy
> workmanship from an anti-freedom commercial organization joined to the hip
> with GCC.
>
> Dave
>
>

Re: GNAT on GCC 10.x has build problems

[Dave Blanchard]

Did you know it's possible to read my post and ruminate upon its meaning without responding this way? I bet you didn't know that.

I'm not asking for any of your 'help', in case it wasn't obvious. I'm quite used to solving the GCC project's problems myself by now, since the GCC project seems determined not to fix its own problems, judging by the typical commonplace response of ingrates such as yourself. I simply asked if this broken piece of shit--which should have been designed correctly decades ago but somehow isn't--has finally been fixed, or if it continues to be a broken piece of shit. 

Dave