[bug#28004] Chromium

[Marius Bakke <mbakke@fastmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1774 bytes --]

ng0 <ng0@n0.is> writes:

> Marius Bakke transcribed 39K bytes:
>
>> Testing and feedback welcome!
>> 
>> Currently there are two "important" (blocking?) TODOs left:
>> 
>> * Move the 'delete-bundled-software' phase to a source snippet.
>>   Repacking the ~500MiB compressed tarball is *really* expensive.  It
>>   should also aid the licensing situation.
>> * Delete the two default entries from the "most used" list on the New
>>   Tab page.  The first run will download thumbnails for these sites,
>>   leaking data.  One of them also leads to the disabled-by-default
>>   store, promoting non-free software.
>> 
>> I'm optimistic that fixing the second item will make the browser not
>> leak *any* data at launch with the default configuration.  Which leads
>> to a third item: writing a system test that verifies that launching
>> Chromium does indeed not initiate any network traffic.
>> 
>> Anyway, here is the latest patch:
>> 
>
>> From f813b2d7ec0728a906720fa74bf9f442af6ab10d Mon Sep 17 00:00:00 2001
>> From: Marius Bakke <mbakke@fastmail.com>
>> Date: Wed, 12 Oct 2016 17:25:05 +0100
>> Subject: [PATCH] gnu: Add chromium.
>> 
>> * gnu/packages/chromium.scm: New file.
>> * gnu/local.mk: Record it.
>
> I think you forgot a package:
>
> gnu/packages/chromium.scm:664:5: icu4c-59.1: unbound variable

Indeed.  This can now be changed to use the regular "icu4c" package.

Tangentially, these kinds of problems are typical with new Chromium
releases.  In 63 or later, system harfbuzz had to be disabled.  If we
are going to carry this package, changes like these *will* be normal.

Upstream only tests their releases with Clang, and with the bundled
versions of packages, regardless of the unbundling script.  Not great.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]