Re: [Mingw-cross-env-list] Issue with distfile names, extraction and variables in download/extraction steps.

[Gregorio Litenstein]

So… I had to update and change a few more packages but I’ve got librsvg (and test-librsvg.exe; although I haven’t actually tested its functionality yet); it’d be really cool if somebody could take a look and try it, I’ve left the previous with the bootstrapping logic up in case we want to do that in the future (although honestly I can’t really see a reason to do it)

https://github.com/Lord-Kamina/mxe/tree/rust-just-download

P.S. Once again, I’ve tested this on gcc 8.3.0; don’t know how 5.5.0 will do.

P.P.S. So far I’ve only tried x86_64-w64-mingw32.shared; will leave a few other targets building in order to see how it goes.

-- 

Gregorio Litenstein Goldzweig 

Médico Cirujano

 

• Fono:  +56 9 96343643
• E-Mail:  address@hidden

On September 15, 2019 at 13:46:54, Gregorio Litenstein (address@hidden) wrote:

We could download everything easily (any version); if we do go the build-all route, I would suggest building all cross-target libstds at once (that would mean hardcoding deps on at least cc for both 64 and 32 bits) but I think that's preferable than 4-5 hours of rebuilding each time. Either that or download.

And so far everybody in the rust discord says this is to be expected because of unstable ABI. 

Since we're mostly using rust as prerequisite for other things, maybe downloading is not a sin (we could stash the building script somewhere in case the situation changes in the future of course)

On Sep 15, 2019, 10:50 -0300, Tony Theodore <address@hidden>, wrote:

On 15 Sep 2019, at 23:39, Gregorio Litenstein <address@hidden> wrote:

1.34 I think but for building purposes there's no point in adding a version older than the latest stable (1.37)

Then we need to build it…the bootstrap python script is monolithic, but more malleable than autotools.

Cheers,

Tony