Re: A few thing we might consider hosting

["Gábor Boskovits" <boskovits@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2218 bytes --]

I was aware of the discussion thread, however I could not find out what the
current state is? Do we plan to modify this behaviour in the next release?

The core-updates thing makes sense, thanks.

2017-12-03 8:19 GMT+01:00 Gábor Boskovits <boskovits@gmail.com>:

> Ok, thanks for clarification.
> I
>
> 2017-12-03 0:29 GMT+01:00 Leo Famulari <leo@famulari.name>:
>
>> On Sat, Dec 02, 2017 at 05:28:51PM +0100, Gábor Boskovits wrote:
>> > Sometimes while working in guix I run into problems because:
>> > 1. a tarball was removed or modified upstream
>> >
>> > It would be great to have the ability to install the latest release in
>> all
>> > the supported ways on all supported architectures, and have the ability
>> do
>> > guix pull without problems.
>> > Last time I tried that it did not worked, because one of the upstream
>> > linux-libre tarballs was removed. It would be nice if we could afford to
>> > host the sources, so that at lesat a bare-bones guixsd suffered no such
>> > problems.
>>
>> We actually do host the sources, but Guix usually tries fetching them
>> from upstream first, which can be annoying. We are discussing this here:
>>
>> <https://bugs.gnu.org/28659>
>>
>> > 2. some packages take very long time to build (notably guile)
>> >
>> > It would be nice, if we could provide the substitutes that the current
>> > core-updates gnu-build-system needs. That would make development that
>> needs
>> > to be done on core-updates much more pleasant to those who are working
>> in a
>> > restricted hardware environment.
>>
>> We use the core-updates branch like this:
>>
>> 1) For a couple months we just push changes to core packages to the
>> branch without worrying about if it works or not.
>> 2) After some time, we try building the branch and fix everything that
>> is broken. Once that is done, we merge it into the master branch, which
>> is what `guix pull` uses by default.
>>
>> So, for most of the life of a core-updates branch, it's likely that no
>> packages will be buildable, and thus we don't even try, so there are no
>> substitutes.
>>
>> Once we start building it, substitutes are available in the normal way.
>>
>
>

[-- Attachment #2: Type: text/html, Size: 3046 bytes --]