Re: Building Matreshka on Windows

["G.B." <bauhaus@futureapps.invalid>]

On 03.11.16 18:50, Simon Wright wrote:


> The fix was (cut back not to bore you)
>
>      <version>
>        <external>${PREFIX}gnatls -v --version</external>
> -      <grep regexp="^GNATLS.+?(\d+(\.\d+)?)" group="1"></grep>
> +      <grep regexp="^GNATLS.+?(\d+)" group="1"></grep>
>      </version>

Two things I find puzzling when I see this setup:

1 - why would compiler writers parse version strings using
     REs that can fail in ways that I'd be tempted to call
      classic when their core business actually depends on
     well working parsers?

2 - what makes compiler writers not use their languages'
     powers for communicating the information between processes
     in a structured way, with ease?

Business angst?


After a few years of seeing version identifiers change in
seemingly arbitrary ways, I am convinced that the technically
relevant information should be communicated using some
protocol to be endorsed by all parties involved.
Making version identifiers reflect hacks rarely works
for all parties.

One of the few micro formats that seem to work without AI is,
I guess, ISO date strings. The "ISO" part should ring a bell
or two.


> I'm very pleased that, since GPRBUILD 2016, I can use e.g.
>
>    for Target use "arm-eabi";
>    for Runtime ("ada") use "ravenscar-sfp-stm32f4";
>
> in the GPR instead of having to specify on the command line.
>



-- 
"HOTDOGS ARE NOT BOOKMARKS"
Springfield Elementary teaching staff