<macc24>
alyssa: i was expecting a 'no' but this is better \o/
<amonakov>
macc24: I think you still can get a 'no' if you ask about Vulkan 1.1 on Midgard %)
<amonakov>
(as opposed to 1.0, that is)
<alyssa>
^ that too
catfella has quit [Remote host closed the connection]
catfella has joined #panfrost
<macc24>
1.0 + VK_EXT_vertex_attribute_divisor, VK_KHR_dedicated_allocation and VK_KHR_maintenance1 is good enough
<chrisf>
macc24: what do you want the first two for specifically?
<macc24>
chrisf: fna3d
stikonas has quit [Read error: Connection reset by peer]
<alyssa>
f'naaaaaaa
<chrisf>
dedicated allocation can be mostly "yeah sure, <drops the extension struct on the floor>"
kaspter has quit [Quit: kaspter]
stikonas has joined #panfrost
<macc24>
D:
* alyssa
is almost through her backlog of administrative work and can even start coding soon ;o
<alyssa>
let's see what awful, awful state I left my fp16 bifrost series in
archetech has joined #panfrost
Ashleee has joined #panfrost
Danct12 has joined #panfrost
davidlt has quit [Ping timeout: 240 seconds]
Danct12 has quit [Ping timeout: 240 seconds]
milkii has left #panfrost ["afe"]
alpernebbi has quit [Quit: alpernebbi]
thecycoone has quit [Quit: Bridge terminating on SIGTERM]
nlhowell has quit [Ping timeout: 245 seconds]
thecycoone has joined #panfrost
<alyssa>
bbrezillon: icecream95: It has come to my attention that driver teams are allowed to drop the requirement to have reviewed-by/acked-by tags in the commit messages themselves.
<alyssa>
(For commits within the driver, I mean. Anything prefixed with pan)
<alyssa>
I have never understood the point of the tags in a post-Marge Bot world... if you want to know who acked a patch (==> who to blame when things were bad), click the Part-of link and read the discussion.
<alyssa>
For both reviewing patches (esp for contributors without push, having to wait for them to amend the commit message or else force-pushing yourself) and for writing patches (the extra cycle of git rebase before every merge), it's been an annoyance ofr me.
<alyssa>
I'm wondering if you all feel similarly? And if so, maybe we should just... not? I thought other Mesa people would have a problem with this but apparently freedreno's dropping the tags and the sky isn't falling there.
<macc24>
alyssa: would that allow random PRs to be merged without reviewed-by?
<alyssa>
macc24: Yes.
<alyssa>
So a lot friendlier for drive-by contributors as well.
<alyssa>
To be clear - I'm not talking about skipping code review. I'll probably still comment r-b (force of habit). I'm specifically talking about the need to stick the tags into the commit messages manually.
<macc24>
then you have approval of one of the least important people here
<macc24>
;D
<alyssa>
macc24: heh
<alyssa>
anholt: ^^
<anholt>
We've been doing this in freedreno for a long time. It's great.
<alyssa>
now back to whatever I was procrastinating on. git rebase, it looks like
<icecream95>
"having to wait for them to amend the commit message". I was using that as a feature, not a bug; often I've caught problems with my MRs while rebasing to add r-bs
<alyssa>
heh
archetech has quit [Quit: Konversation terminated!]
<italove>
alyssa: sorry, I hadn't read your message here about MR 10067, but I guess we're doing the "no r-b" tags now?
<italove>
"no r-b tags" *
<alyssa>
italove: I mean I can unassign it to Marge if you feel like adding my tag :p
<italove>
ok, I'll do that
WoC has joined #panfrost
<italove>
alyssa: I think you don't have to unassign it, she's late :P
<alyssa>
margieee
<alyssa>
Register preloading bug wee.
<alyssa>
ugh I need to rebase on top of my RA rewrite branch
<alyssa>
um.. one thing at a time partner
<alyssa>
I have too many fricking patches
* alyssa
tries to whittle down set
<HdkR>
unordered_set to the rescue
warpme_ has quit [Quit: Connection closed for inactivity]
<chrisf>
*groan*
<alyssa>
groan?
<HdkR>
Bad C++ references make chrisf groan. Should have used an r-value reference instead
<HdkR>
&&groan
<icecream95>
alyssa: What does DCD stand for?
zkrx has quit [Ping timeout: 246 seconds]
<alyssa>
icecream95: Draw call descriptor ("Draw" in the XML)
stikonas has quit [Remote host closed the connection]
<urjaman>
and here i was thinking Data Carrier Detect
stikonas has joined #panfrost
zkrx has joined #panfrost
<alyssa>
The XML uses less... jargon-y terms where possible
<alyssa>
"Draw" instead of DCD, "Renderer" instead of RSD, "Viewport" instead of VPD, "Framebuffer" instead of FBD