Thanks, applied :)

On 2/13/23 17:14, psykose wrote:
> from https://www.freedesktop.org/software/appstream/docs/chap-Metadata.html#spec-component-location :
>
>   Upstream projects can ship one or more metainfo files in /usr/share/
>   metainfo/%{id}.metainfo.xml,
> ---
>
> apparently reading is hard- the filename gets renamed too.
>
>   data/meson.build                                            | 6 +++---
>   ...appdata.xml.in => nl.brixit.Thumbdrives.metainfo.xml.in} | 0
>   2 files changed, 3 insertions(+), 3 deletions(-)

Thanks, applied :)

On 2/2/23 16:42, ~lihis wrote:
> Appstream data validation fails with "appstream-util validate-relax
> --nonet" as there is two <release>'s with same version. This patch
> removes the duplicate entry thus making the validation to pass.
>
> Tomi Lähteenmäki (1):
>    Remove duplicate version from appstream data
>
>   data/org.postmarketos.Numberstation.appdata.xml | 10 +---------
>   1 file changed, 1 insertion(+), 9 deletions(-)
>

Hi Andrey,

Sorry for the delay, I've applied your patches.
I had not encountered otp urls with decimals in them before, they don't 
really make sense for things like length but better to handle it than 
not I guess.

Greetings,
Martijn Braam

On 1/14/23 11:33, Andrey Skvortsov wrote:
> Hi Martijn,
>
> Do you have any feedback on these patches sent a month ago?

Thanks, applied :)

On 2/22/23 17:15, Evangelos Ribeiro Tzaras wrote:
> The icon was introduced in
> ffa0b6a57ebed3f851d79723862c82bc927ff732
> but never actually used (user-trash-symbolic is used instead).
> ---
>   numberstation/icon/delete-symbolic.svg    | 1 -
>   numberstation/numberstation.gresource.xml | 2 --
>   2 files changed, 3 deletions(-)
>   delete mode 100644 numberstation/icon/delete-symbolic.svg
>
> diff --git a/numberstation/icon/delete-symbolic.svg b/numberstation/icon/delete-symbolic.svg
> deleted file mode 100644

Thanks, applied :)

On 2/22/23 11:15, Hugo Osvaldo Barrera wrote:
> Many projects add a leading `v` to tags for versions. However, the
> contents of tarballs don't include this, so `$builddir` ends up being
> `package-v1.0.0`, but the real path is actually `package-1.0.0`.
>
> I haven't seem any scenarios where a leading `v` is present but this
> doesn't uphold.
>
> That aside, APKBUILD conventions seem to align with removing the leading
> `v`.
> ---
>   createaport/forge.py | 4 ++++

It's possible to set the default To: address and subject prefix in the
git config of the local checkout. This makes the workflow a bit easier
and makes sure the subject starts with [PATCH pmbootstrap] instead of
the regular [PATCH].
---
 README.md | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/README.md b/README.md
index 005e7bc1..de3554f3 100644
--- a/README.md
+++ b/README.md
@@ -17,6 +17,12 @@ Send patches via mail or web UI to
 ~postmarketos/pmbootstrap-devel@lists.sr.ht
[message trimmed]

This behavior is because it's title-casing it with a few exceptions.

https://git.sr.ht/~martijnbraam/keyring/tree/master/item/himitsu_gtk/window.py#L83

I did not account for imaps vs imap here, mainly there's not really a 
spec for this part. The title casing works a lot better with everything 
that's not an acronym like the web protocol.

On 10/22/22 01:59, Hugo Osvaldo Barrera wrote:
> For some odd reason, the casing of the protocol "titles" (e.g.: the labels on
> the top of each protocol) is mixed up; the first one is shown all uppercase
> ("SMTP"), while others are title-case ("Imaps", "Smtps", etc).
>
> A screenshot might be much clearer, but emails with image/png attachment are

On 7/6/22 15:06, Caleb Connolly wrote:
>
>
> On 05/07/2022 17:10, Martijn Braam wrote:
>> Make the types for variables more consistent to limit casts. All
>> physical dimensions are floats and pixel coordinates are long except on
>> the interface with tflib which uses int.
> Out of interest, why use long for pixel coordinates instead of int? Do 
> any values get close to exceeding INT_MAX?

they don't, but on ARM int and long are both 32 bits. since most C apis 
dealing with numbers give long as result making all
the coordinates consistently long makes most of the code a bit neater. 
the (int)somelong casts are no-ops on the ARM builds when checking with

Make the types for variables more consistent to limit casts. All
physical dimensions are floats and pixel coordinates are long except on
the interface with tflib which uses int.
---
 include/pbsplash.h |  2 +-
 src/animate.c      | 30 +++++++++------
 src/pbsplash.c     | 96 ++++++++++++++++++++++------------------------
 3 files changed, 65 insertions(+), 63 deletions(-)

diff --git a/include/pbsplash.h b/include/pbsplash.h
index b855f28..423f128 100644
--- a/include/pbsplash.h
+++ b/include/pbsplash.h
@@ -10,6 +10,6 @@ struct col {
[message trimmed]

There are boot flow mockups on 
https://gitlab.com/postmarketOS/pmaports/-/issues/1356

My proposal:

1. Add an unix socket to pbsplash that is listened on for commands while 
it's displaying the splash screen in the initfs

2. Create a pbctl utility to send messages to that socket to update the 
screen

There would be 4 different states for the UI that can be switched to 
with pbctl: