[Replicant] Changes for 6.0 0004 release

[Joonas Kylmälä]

Hi,

Denis 'GNUtoo' Carikli:
> On my side I should improve reviews time. I should also look into
> installing a patchwork instance to keep track of the patches I forgot
> about.

I have an image in my mind that maintaining patchwork for 58 or so git
repositories was difficult. If it doesn't turn out well there are other
options that we can try. Google was using Gerrit at least at some point
for AOSP development so we can look also if that is an option or if
there is some newly adopted patch review system for AOSP or LineageOS.

> 
>> [0001-Adapt-to-recent-LineageOS-forks-LineageOS-mirror.patch
>> [0001-Adapt-to-recent-LineageOS-forks-replicant-6.0-dev.patch
> Both commit messages don't give much details about why this is needed,
> but in another hand it might take some time to understand why upstream
> (lineageOS) might have done that too.
> Acked-By: Denis 'GNUtoo' Carikli <GNUtoo at no-log.org>

You have to read the commits mentioned in the dev branch's commit
message to know why. I think from Replicant project's point of view it
is enough to know that we have "adapted" the manifest to be same as in
upstream. In ideal world I would just cherry-pick the LineageOS commit
but the manifest repositories have diverged too much because our
repository names start with "replicant" and LineageOS ones start with
"LineageOS" so if there are some reorderings in the manifest it just
creates a big git merge conflict.

Though, maybe I could have explained that we want to stay as similar as
possible to LineageOS.

> 
>> [0001-Revert-Point-ResXMLTree-to-a-copy-of-the-asset-buffe.patch
> The commit reverted by the patch above is this one:
>> commit 953d2a3493aaceb3209afc7eef1ff1e7f7242908
>> (replicant/replicant-6.0, m/replicant-6.0-dev) Author: Wolfgang
>> Wiedmeyer <wolfgit at wiedmeyer.de> Date:   Thu Nov 2 23:01:34 2017 +0100
>>
>>     Point ResXMLTree to a copy of the asset buffer in
>> AssetManager::getPkgName() 
>>     This ensures that there is no dangling pointer when the asset is
>> closed. Without this fix, a device ends up in a boot loop when an app
>> is installed that was created using aapt2.
>>     
>>     Reference:
>>     https://github.com/TwidereProject/Twidere-Android/issues/963#issuecomment-335616312
>>     
>>     Signed-off-by: Wolfgang Wiedmeyer <wolfgit at wiedmeyer.de>
> 
> Your commit message could give a bit more context on the issue, for
> instance with something like that:
>> This reverts commit 953d2a3493aaceb3209afc7eef1ff1e7f7242908.
>>
>> The 953d2a3493aaceb3209afc7eef1ff1e7f7242908 commit (Point ResXMLTree
>> to a copy of the asset buffer in AssetManager::getPkgName()) was
>> a fix for preventing devices from ending in a boot loop when an
>> application that was created with aapt2 was installed.
>>
>> Since upstream has now fixed it we can now drop this patch to use
>> upstream's fix instead. 

Will fix this.

> 
>> [0002-Update-F-Droid-Privileged-Extension-to-0.2.8-replicant-6.0-dev.patch
> This looks good, keeping F-Droid privileged extension doesn't harm.
> Acked-By: Denis 'GNUtoo' Carikli <GNUtoo at no-log.org>
> 
> Note that I didn't have the ability to test the patches (I was at a
> conference with limited time to do remote builds).

You would also need to test the patches together with the
replicant-6.0-dev branch's manifest that can be found from
git://git.putti.eu/replicant/manifest.git because some of these patches
depend on the other merged repositories.

Joonas

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.osuosl.org/pipermail/replicant/attachments/20180915/5e334701/attachment.asc>