@hermmm Alright so at least on `gnome-session` it only leads to a silent syslog entry: `WARNING: De

@hermmm Alright so at least on

gnome-session

gnome-session

gnome-session

gnome-session it only leads to a silent syslog entry:

WARNING: Desktop file /var/home/johnny/.config/autostart/foo.desktop for application foo.desktop could not be parsed or references a missing TryExec binary

WARNING: Desktop file /var/home/johnny/.config/autostart/foo.desktop for application foo.desktop could not be parsed or references a missing TryExec binary

WARNING: Desktop file /var/home/johnny/.config/autostart/foo.desktop for application foo.desktop could not be parsed or references a missing TryExec binary

WARNING: Desktop file /var/home/johnny/.config/autostart/foo.desktop for application foo.desktop could not be parsed or references a missing TryExec binary

I'll try KDE too. If both are fine with missing links, I'd say it's tested enough to be okay leaving an orphan?

Edit: KDE gives a silent syslog warning too, from

systemd-xdg-aut

systemd-xdg-aut

systemd-xdg-aut

systemd-xdg-aut which says

/var/home/johnny/.config/autostart/foo.desktop: stat() failed, ignoring: No such file or directory

/var/home/johnny/.config/autostart/foo.desktop: stat() failed, ignoring: No such file or directory

/var/home/johnny/.config/autostart/foo.desktop: stat() failed, ignoring: No such file or directory

/var/home/johnny/.config/autostart/foo.desktop: stat() failed, ignoring: No such file or directory

bsherman•5/12/23, 5:50 PM

@Arcitec I'll add this to the paths proposal discussion after we chat but... how passionate are you about the "no clunky paths"?

I'm looking at how to move from /usr/share/ublue-os/ublue-os-update-services/usr/share/ublue-os/ublue-os-update-services to /usr/share/ublue-os/update-services/usr/share/ublue-os/update-services and I don't like what I'm seeing.

It's due to the RPM spec, and I really like the cleanliness of the spec as is.

See this line for example: https://github.com/ublue-os/config/blob/main/rpmspec/ublue-os-update-services.spec#L49

%attr(0644,root,root) %{_datadir}/%{VENDOR}/%{NAME}/%{_sysconfdir}/rpm-ostreed.conf

%attr(0644,root,root) %{_datadir}/%{VENDOR}/%{NAME}/%{_sysconfdir}/rpm-ostreed.conf

The %{VENDOR}%{VENDOR} is ublue-osublue-os and the %{NAME}%{NAME} is ublue-os-update-servicesublue-os-update-services.

edit: add local context

AArcitec @hermmm Alright so at least on `gnome-session` it only leads to a silent syslog ...

xyny•5/12/23, 5:50 PM

Sure!

Bbsherman @Arcitec I'll add this to the paths proposal discussion after we chat but... how...

bsherman•5/12/23, 5:52 PM

I don't want the re-name the RPM as it should be ublue-os-update-servicesublue-os-update-services in the RPM database.

Bbsherman @Arcitec I'll add this to the paths proposal discussion after we chat but... how...

ArcitecOP•5/12/23, 5:52 PM

Ew. I see the point but I think there's a way in RPM specs to create dynamic variables, where we could add one that's something like %SHARED_DATA_PATH=${NAME:/-ublue-os/}%SHARED_DATA_PATH=${NAME:/-ublue-os/} (pseudocode)

ArcitecOP•5/12/23, 5:52 PM

I remember seeing people do something dynamic like that to get the first 7 characters of Git commit hashes

AArcitec Ew. I see the point but I think there's a way in RPM specs to create dynamic var...

bsherman•5/12/23, 5:53 PM

rpmspec is not my strength, I'm definitely learning

ArcitecOP•5/12/23, 5:54 PM

There is some way, similar to bash variable substitution. So I think it's worth trying to solve that. Dynamically cutting away the ublue-os-ublue-os- prefix at that location. We should avoid the redundancy, and we're referring to the full path in other places such as mainmain.

bsherman•5/12/23, 5:54 PM

yeah, I'll read up https://rpm-packaging-guide.github.io/

ArcitecOP•5/12/23, 5:55 PM

I will try to find a spec that does it.

ArcitecOP•5/12/23, 5:55 PM

lol nvm, COPR is down https://copr-dist-git.fedorainfracloud.org/cgit/sunwire/virt-manager/virt-manager.git

ArcitecOP•5/12/23, 5:55 PM

That spec does it but I can't access it :/

bsherman•5/12/23, 5:55 PM

I like the idea of maybe having a "UBLUE_SUB_NAME" which conditionally removes any "ublue-os-" prefix from "%{NAME}" otherwise uses %{NAME}

ArcitecOP•5/12/23, 5:56 PM

Yeah and if we can have a nice pattern to auto-generate it, we'll be set for all rpmspecs

bsherman•5/12/23, 5:58 PM

i do want to make the point... this is basically vanity, though

ArcitecOP•5/12/23, 5:59 PM

It is but it's such nice and fresh vanity / bling.

ArcitecOP•5/12/23, 6:00 PM

I am searching too, to help out.

bsherman•5/12/23, 6:00 PM

appreciate it

ArcitecOP•5/12/23, 6:00 PM

I found it

bsherman•5/12/23, 6:01 PM

if reasonably easy, and still clean/templated like our specs are, i'm happy to do it... but if it's super jank or becomes a lot of hard coding, i would push back... reasonable?

ArcitecOP•5/12/23, 6:01 PM

https://rpm-software-management.github.io/rpm/manual/macros.html

%define <name>[(opts)] <body>%define <name>[(opts)] <body>

With this function:

%{gsub:...}    replace occurences of pattern in a string
                        (see Lua `string.gsub()`)

%{gsub:...}    replace occurences of pattern in a string
                        (see Lua `string.gsub()`)

rpm.org - Macro syntax

bsherman•5/12/23, 6:01 PM

ah, it's the %define

ArcitecOP•5/12/23, 6:01 PM

I don't know exactly how to use it though, but that's the one. It does a string substitute and only changes if it finds it, else it uses the original.

bsherman•5/12/23, 6:01 PM

ok, i'll play with this

ArcitecOP•5/12/23, 6:01 PM

Since it's a LUA pattern, this would do for the pattern: ^ublue-os-^ublue-os-

bsherman•5/12/23, 6:02 PM

i think i just hit that in other docs source

ArcitecOP•5/12/23, 6:02 PM

yea

ArcitecOP•5/12/23, 6:03 PM

@bsherman

string.gsub("ublue-os-startup-scripts", "^ublue%-os%-", "")
startup-scripts    1

string.gsub("ublue-os-startup-scripts", "^ublue%-os%-", "")
startup-scripts    1

ArcitecOP•5/12/23, 6:04 PM

That's the exact pattern we'll need to use, since lua interprets plain dashes as character ranges (lol).

ArcitecOP•5/12/23, 6:04 PM

"^ublue%-os%-""^ublue%-os%-"

ArcitecOP•5/12/23, 6:07 PM

After reading a bit of the macro examples, I assume that it would look similar to this in use:

%define project_name %{gsub:"^ublue%-os%-", ""}%define project_name %{gsub:"^ublue%-os%-", ""}

bsherman•5/12/23, 6:08 PM

thanks, i'm looking at other changes I need to make related to this, so appreciate you digging on this part

ArcitecOP•5/12/23, 6:09 PM

No problem. I wish that COPR was up so I could see the exact code used by some repos I know that use this, lol.

bsherman•5/12/23, 6:09 PM

i think it's up its just they did changes to git hosting, no?

ArcitecOP•5/12/23, 6:10 PM

bsherman•5/12/23, 6:10 PM

https://copr.fedorainfracloud.org/coprs/sunwire/virt-manager/

ArcitecOP•5/12/23, 6:10 PM

Google is a genius!

ArcitecOP•5/12/23, 6:10 PM

"gsub" filetype:spec"gsub" filetype:spec Google that

ArcitecOP•5/12/23, 6:10 PM

I found an example that uses it https://src.fedoraproject.org/rpms/sil-charis-fonts/blob/f38/f/sil-charis-fonts.spec

ArcitecOP•5/12/23, 6:10 PM

bsherman•5/12/23, 6:11 PM

very nice

ArcitecOP•5/12/23, 6:11 PM

Interestingly they use the full "run arbitrary lua" method for it, I'll see if I can find one that uses the shorthand for just doing gsub.

bsherman•5/12/23, 6:12 PM

from what i'm reading it's still run in the rpm-embedded lua, not a spawn/fork

ArcitecOP•5/12/23, 6:13 PM

Yeah, it's probably running with eval()eval() or equivalent. No big deal. I just wanna know how to use the neater shorthand so we don't need to

print()

print()

the result of gsub manually.

bsherman•5/12/23, 6:15 PM

this isn't it? %define project_name %{gsub:"^ublue%-os%-", ""}%define project_name %{gsub:"^ublue%-os%-", ""}

Bbsherman https://copr.fedorainfracloud.org/coprs/sunwire/virt-manager/

ArcitecOP•5/12/23, 6:15 PM

Oh and that link is just the copr page, but I need to look at the git repo for it, which is in the packages tab. That's what's broken

AArcitec Oh and that link is just the copr page, but I need to look at the git repo for i...

bsherman•5/12/23, 6:15 PM

right, i think they actually shutdown the git

Bbsherman this isn't it? `%define project_name %{gsub:"^ublue%-os%-", ""}`

ArcitecOP•5/12/23, 6:16 PM

That's my best guess, I can't try it though, no mockmock on this system and not much knowledge about how to run it.

bsherman•5/12/23, 6:16 PM

no worries, i'm working on a local test build

ArcitecOP•5/12/23, 6:16 PM

Nice

And worst case scenario we can always use the longform that fedora uses in the repo I found hehe.

ArcitecOP•5/12/23, 6:20 PM

Interesting. https://rpm-software-management.github.io/rpm/manual/macros.html The "Defining a macro" section explains that ${gsub:...}${gsub:...} is a macro, and explains that macros can take multiple arguments. %1, %2 etc.

Next hint is that the gsubgsub macro just forwards the call to Lua's string.gsub:

%{gsub:...}    replace occurences of pattern in a string
                        (see Lua `string.gsub()`)

%{gsub:...}    replace occurences of pattern in a string
                        (see Lua `string.gsub()`)

So my updated guess would be the following to line up with all arguments taken by string.gsub:

%global project_name %{gsub:"%{NAME}", "^ublue%-os%-", ""}%global project_name %{gsub:"%{NAME}", "^ublue%-os%-", ""}

Possibly without the "" surrounding %{NAME}%{NAME} in case that halts expansion. But it seems correct to use quotes like that, judging by:

https://src.fedoraproject.org/rpms/sil-charis-fonts/blob/f38/f/sil-charis-fonts.spec

Edit: Oh use %global%global not %define%define, the latter creates macros!