Hi there,
I'm currently working on the PatchTracking[1] spec and, wrt patch series, I'm wondering whether you consider the patches that are part of a series to be separate or the series to be a single entity?
Thanks!
[1] https://wiki.linaro.org/Platform/Infrastructure/Specs/PatchTracking
On 18 March 2011 23:29, Guilherme Salgado guilherme.salgado@linaro.org wrote:
I'm currently working on the PatchTracking[1] spec and, wrt patch series, I'm wondering whether you consider the patches that are part of a series to be separate or the series to be a single entity?
Depends. By putting a patchset into a series I definitely mean that in some ways it should be considered a single entity -- so for instance in my list of pending qemu patches a series is a single list entry, and generally they're semantically related changes intended to be all applied together. On the other hand sometimes you do want to think about the patches separately -- sometimes patches 1,4,5 get reviewed OK and 2,3 rejected for rework; occasionally a partial patchset is committed.
So: a series is a single entity containing a lot of separate patches; put another way, what's the purpose for which you want to decide whether it's one thing or a whole pile? :-)
-- PMM
On Mon, 2011-03-21 at 18:24 +0000, Peter Maydell wrote:
On 18 March 2011 23:29, Guilherme Salgado guilherme.salgado@linaro.org wrote:
I'm currently working on the PatchTracking[1] spec and, wrt patch series, I'm wondering whether you consider the patches that are part of a series to be separate or the series to be a single entity?
Depends. By putting a patchset into a series I definitely mean that in some ways it should be considered a single entity -- so for instance in my list of pending qemu patches a series is a single list entry, and generally they're semantically related changes intended to be all applied together. On the other hand sometimes you do want to think about the patches separately -- sometimes patches 1,4,5 get reviewed OK and 2,3 rejected for rework; occasionally a partial patchset is committed.
Fair enough.
So: a series is a single entity containing a lot of separate patches; put another way, what's the purpose for which you want to decide whether it's one thing or a whole pile? :-)
That's to decide whether we should count the patches that belong to a series individually or not.
On Tue, Mar 22, 2011 at 4:05 PM, Guilherme Salgado guilherme.salgado@linaro.org wrote:
On Mon, 2011-03-21 at 18:24 +0000, Peter Maydell wrote:
On 18 March 2011 23:29, Guilherme Salgado guilherme.salgado@linaro.org wrote:
I'm currently working on the PatchTracking[1] spec and, wrt patch series, I'm wondering whether you consider the patches that are part of a series to be separate or the series to be a single entity?
Depends. By putting a patchset into a series I definitely mean that in some ways it should be considered a single entity -- so for instance in my list of pending qemu patches a series is a single list entry, and generally they're semantically related changes intended to be all applied together. On the other hand sometimes you do want to think about the patches separately -- sometimes patches 1,4,5 get reviewed OK and 2,3 rejected for rework; occasionally a partial patchset is committed.
Fair enough.
So: a series is a single entity containing a lot of separate patches; put another way, what's the purpose for which you want to decide whether it's one thing or a whole pile? :-)
That's to decide whether we should count the patches that belong to a series individually or not.
I don't have a strong opinion on this, but given that designing a decent patch series, reworking them, splitting them up etc. is more work than committing everything from top i don't see why the patches shouldn't be counted.
Also if other folks run statistic on contributions to lets say kernel.org they typically just count patches/commits and not series afaiui.
On 03/22/2011 08:05 AM, Guilherme Salgado wrote:
That's to decide whether we should count the patches that belong to a series individually or not.
Could you generate two sets of stats like "Total Patches Per Author" and "Total Patch-Series Per Author" or possibly try and give more weight to a patch series based on the number of individual patches?
Trying to count say my one-line patch to fix a build break in x-loader to a patch series like "device tree support" doesn't seem fair.
-andy
On 22 March 2011 16:56, Andy Doan andy.doan@linaro.org wrote:
On 03/22/2011 08:05 AM, Guilherme Salgado wrote:
That's to decide whether we should count the patches that belong to a series individually or not.
Could you generate two sets of stats like "Total Patches Per Author" and "Total Patch-Series Per Author" or possibly try and give more weight to a patch series based on the number of individual patches?
Trying to count say my one-line patch to fix a build break in x-loader to a patch series like "device tree support" doesn't seem fair.
We could count number of lines changed as well :-)
-- PMM
On Tue, Mar 22, 2011 at 11:10 AM, Peter Maydell peter.maydell@linaro.org wrote:
On 22 March 2011 16:56, Andy Doan andy.doan@linaro.org wrote:
On 03/22/2011 08:05 AM, Guilherme Salgado wrote:
That's to decide whether we should count the patches that belong to a series individually or not.
Could you generate two sets of stats like "Total Patches Per Author" and "Total Patch-Series Per Author" or possibly try and give more weight to a patch series based on the number of individual patches?
Trying to count say my one-line patch to fix a build break in x-loader to a patch series like "device tree support" doesn't seem fair.
We could count number of lines changed as well :-)
I believe the scripts that Greg KH and Jonathan Corbet use for kernel development tracking do that. http://www.kernel.org/pub/linux/kernel/people/gregkh/kernel_history/
-- PMM
linaro-dev mailing list linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-dev
On Tue, 2011-03-22 at 09:56 -0700, Andy Doan wrote:
On 03/22/2011 08:05 AM, Guilherme Salgado wrote:
That's to decide whether we should count the patches that belong to a series individually or not.
Could you generate two sets of stats like "Total Patches Per Author" and "Total Patch-Series Per Author" or possibly try and give more weight to a patch series based on the number of individual patches?
Trying to count say my one-line patch to fix a build break in x-loader to a patch series like "device tree support" doesn't seem fair.
Yeah, I think we'll want to do that (as well as counting lines changed/added/removed in the patches themselves) at some point, but right now I think we're most interested in checking that patches are flowing upstream, so to start with we'll be counting individual patches.
Cheers,