Skip to contentSkip to conversation

DPDK patch backlog

DISCUSSION Stephen Hemminger 8 participants 14 messages Oct 15, 2015 – Oct 23, 2015

lore
Share & tools
lore
#0 lore raw

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?
The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

#1 lore raw

+1

-----Original Message-----
From: dev [mailto:[email protected]] On Behalf Of Stephen Hemminger
Sent: Friday, October 16, 2015 5:44 AM
To: Thomas Monjalon <[email protected]>
Cc: [email protected]
Subject: [dpdk-dev] DPDK patch backlog

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?
The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

#3 lore raw
In reply to Stephen Hemminger

2015-10-15 14:44, Stephen Hemminger:

Quoted 3 lines

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?

Yes

Quoted 3 lines

The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

That's why having a drivers/net subtree would be useful.

#2 lore raw
In reply to Zhu, Heqing

Hi Hemminger,
+1

Quoted 16 lines

-----Original Message-----
From: dev [mailto:[email protected]] On Behalf Of Zhu, Heqing
Sent: Friday, October 16, 2015 10:47 AM
To: Stephen Hemminger; Thomas Monjalon
Cc: [email protected]
Subject: Re: [dpdk-dev] DPDK patch backlog

+1

-----Original Message-----
From: dev [mailto:[email protected]] On Behalf Of Stephen Hemminger
Sent: Friday, October 16, 2015 5:44 AM
To: Thomas Monjalon <[email protected]>
Cc: [email protected]
Subject: [dpdk-dev] DPDK patch backlog

Quoted 5 lines

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?
The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

I think Thomas still needs some reason. Would you like to ack my patches for e1000 base driver :)

#4 lore raw
In reply to Thomas Monjalon
Quoted 5 lines · Thomas Monjalon

2015-10-15 14:44, Stephen Hemminger:

Quoted 3 lines

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?

Quoted 8 lines

Yes

Quoted 3 lines

The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

That's why having a drivers/net subtree would be useful.

Agreed, a dpdk-next tree would really be the solution here.

Neil

#7 lore raw

2015-10-20 21:34, Stephen Hemminger:

Quoted 4 lines

Patch backlog is not getting better, now at 486.

How can we break this logjam?
Do I need to make a new "ready for merge" tree?

What would mean "ready for merge"?
A lot of patches are acked but do not compile or doc is missing.

I have the feeling it would become easier if there were mailing-list, patchwork and git tree dedicated to some areas. What about starting with drivers/net?

#8 lore raw
Quoted 8 lines · Thomas Monjalon

2015-10-20 21:34, Stephen Hemminger:

Quoted 4 lines

Patch backlog is not getting better, now at 486.

How can we break this logjam?
Do I need to make a new "ready for merge" tree?

What would mean "ready for merge"?
A lot of patches are acked but do not compile or doc is missing.

Well, isn't that one quite reasonable definition of being "ready"?
- patch must be acked
- patch must apply and compile (when relevant)
- is appropriately documented (commit message style and all)

- Panu -

Quoted 4 lines

I have the feeling it would become easier if there were mailing-list, patchwork and git tree dedicated to some areas. What about starting with drivers/net?

#9 lore raw

2015-10-21 11:48, Panu Matilainen:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

2015-10-20 21:34, Stephen Hemminger:

Quoted 4 lines

Patch backlog is not getting better, now at 486.

How can we break this logjam?
Do I need to make a new "ready for merge" tree?

Quoted 7 lines
Quoted 2 lines

What would mean "ready for merge"?
A lot of patches are acked but do not compile or doc is missing.

Well, isn't that one quite reasonable definition of being "ready"?
- patch must be acked
- patch must apply and compile (when relevant)
- is appropriately documented (commit message style and all)

Yes.
Compilation must be tested with GCC and clang, as static and shared libraries and for 32-bit and 64-bit targets.
Documented means good commit message and doc or release notes updated.

#13 lore raw
In reply to Thomas Monjalon
Quoted 8 lines · Thomas Monjalon

2015-10-20 21:34, Stephen Hemminger:

Quoted 4 lines

Patch backlog is not getting better, now at 486.

How can we break this logjam?
Do I need to make a new "ready for merge" tree?

What would mean "ready for merge"?
A lot of patches are acked but do not compile or doc is missing.

I would think that many of these patches that don't compile should be marked as rejected in patchwork, and a note sent to the list. If you can enable the patchwork xmlrpc interface, we could script/automate that process

Neil

Quoted 5 lines

I have the feeling it would become easier if there were mailing-list, patchwork and git tree dedicated to some areas. What about starting with drivers/net?

#10 lore raw
In reply to Thomas Monjalon
Quoted 2 lines · Thomas Monjalon

Compilation must be tested with GCC and clang, as static and shared libraries and for 32-bit and 64-bit targets.

Is this process scripted somewhere?

Matthew.

#5 lore raw
In reply to Neil Horman
Quoted 5 lines · Neil Horman
Quoted 4 lines · Thomas Monjalon

2015-10-15 14:44, Stephen Hemminger:

Quoted 3 lines

There are currently 428 patches in New state in DPDK patchwork.

Thomas, could you start reducing that backlog?

Quoted 8 lines
Quoted 7 lines

Yes

Quoted 3 lines

The simplest solution would be to merge some of the big patch series from Intel for the base drivers, then reviewers can focus on the other patches.

That's why having a drivers/net subtree would be useful.

Agreed, a dpdk-next tree would really be the solution here.

Can't agree more :)

Thanks,
Michael

Quoted 3 lines

Neil

#12 lore raw
In reply to Thomas Monjalon
Quoted 9 lines · Thomas Monjalon

2015-10-21 11:48, Panu Matilainen:

Quoted 8 lines · Thomas Monjalon
Quoted 7 lines

2015-10-20 21:34, Stephen Hemminger:

Quoted 4 lines

Patch backlog is not getting better, now at 486.

How can we break this logjam?
Do I need to make a new "ready for merge" tree?

What would mean "ready for merge"?
A lot of patches are acked but do not compile or doc is missing.

Quoted 8 lines
Quoted 4 lines

Well, isn't that one quite reasonable definition of being "ready"?
- patch must be acked
- patch must apply and compile (when relevant)
- is appropriately documented (commit message style and all)

Yes.
Compilation must be tested with GCC and clang, as static and shared libraries and for 32-bit and 64-bit targets.
Documented means good commit message and doc or release notes updated.

What about bug fix patches?

Thanks,
Michael

Quoted 1 line