Re: [webkit-dev] Changes to the WebKit2 development process
IMHO, recently, contributors as well as WK2 owners are a bit suffering from it. I think we can go to a better direction: making all developers happy as well as resolving WK2 owners' concerns. I think the straightforward way to achieve this is to let somebody be an owner from every port. He/she should could be restricted to approve only platform specific changes. ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
El mar, 29-01-2013 a las 09:39 +0100, Balazs Kelemen escribió: IMHO, recently, contributors as well as WK2 owners are a bit suffering from it. I think we can go to a better direction: making all developers happy as well as resolving WK2 owners' concerns. I think the straightforward way to achieve this is to let somebody be an owner from every port. He/she should could be restricted to approve only platform specific changes. And why not making it even simpler an allow any reviewer to r+ port-specific patches? If the patch doesn't change any cross-platform code at all there's no risk to break mac or any other port. Carlos Garcia Campos http://pgp.rediris.es:11371/pks/lookup?op=getsearch=0xF3D322D0EC4582C3 ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
Hi, Carlos Garcia Campos írta: El mar, 29-01-2013 a las 09:39 +0100, Balazs Kelemen escribió: IMHO, recently, contributors as well as WK2 owners are a bit suffering from it. I think we can go to a better direction: making all developers happy as well as resolving WK2 owners' concerns. I think the straightforward way to achieve this is to let somebody be an owner from every port. He/she should could be restricted to approve only platform specific changes. And why not making it even simpler an allow any reviewer to r+ port-specific patches? If the patch doesn't change any cross-platform code at all there's no risk to break mac or any other port. The answer is simple: Because Apple decided and proclaimed this fiat ... br, Ossy ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
2013/1/10 Sam Weinig wei...@apple.com At this point, we ask that all completely non-trivial patches be reviewed by an owner, even if in port specific code. - Sam Hi Sam and webkit developers. I really appreciate WK2 owners' effort to improve all port specific code as well as WK2 code. However, recently EFL, Gtk and Qt (by Alphabetical order) developers are having a hard time due to decreasing productivity. For example, I had filed a platform specific patch and had it reviewed by a Qt reviewer 10 days ago. And then WK2 owner's review was needed to commit. One of the WK2 owners made an effort to review. However, compared to before, review speed dramatically decreased and I had to wait his response for several days. I understand the WK2 owner had many jobs besides reviewing this platform specific patch, and I really appreciated his effort and comment. Finally I had to separate the patch to a WebCore platform specific part and a WK2 platform specific part because many other bugs were blocked by the WebCore part. IMHO, recently, contributors as well as WK2 owners are a bit suffering from it. I think we can go to a better direction: making all developers happy as well as resolving WK2 owners' concerns. Cheers, Dongsung Huang ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
On Tuesday, January 08, 2013 02:57:53 PM Sam Weinig wrote: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. I think the fact that the regular WebKit review process stops at the boundary of WebKit2 should be documented in the WebKit Committers and Reviewer Policy. Simon ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
On Wed, Jan 9, 2013 at 11:32 AM, Simon Hausmann simon.hausm...@digia.com wrote: On Tuesday, January 08, 2013 02:57:53 PM Sam Weinig wrote: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. I think the fact that the regular WebKit review process stops at the boundary of WebKit2 should be documented in the WebKit Committers and Reviewer Policy. Agree. And please clarify on the policy if we are talking about everything inside the WebKit2/ directory or if we have exceptions. It is not clear to me if port specific code is covered by this rule and should by reviewed by the owners. And what about code shared by Qt, GTK and EFL (i.e. Platform/CoreIPC/unix/) but not used by Mac? - Thiago ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
Hi, 2013/1/8 Sam Weinig wei...@apple.com: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. Isn't that why we already differentiate between committers and reviewers? I mean, isn't like that throughout the entire project already? I thought _any_ patch to any part of WebKit required significant expertise to be evaluated. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. If I'm not mistaken, there are only people from the Mac port in the OWNERS file. Will there be some policy that other reviewers from other ports can become owners of WebKit2 as well, or will that be Apple-only always? 2) Ports must keep themselves building. Non Apple Mac ports, if broken by core functionality changes to WebKit2, are now responsible for fixing themselves. We have asked those who run the EWS bots to make sure that failing to build WebKit2 does not block the commit queue from committing. IMHO, doing this is breaking down an entire 'culture' of the WebKit workflow that we are all so proud of. 3) Over time, owners may remove peripheral functionality from the main WebKit2 directory, such as support for features that aren't broadly applicable. We will not do this immediately, and we will work with ports that are interested in such features to create appropriate, maintainable general-purpose mechanisms that can be used to implement them outside of core WebKit2 code. While we understand that this change will inconvenience some ports, we have decided that forward progress of WebKit2 is a more important concern, and we are moving forward with this change tonight. Well, at least from my side, I only got this email _after_ you had already moved forward with everything. I actually saw the patches landing way before it. Not cool! :) I thought the reviewers had all agreed about all these, but now after the first round of replies to this thread it is sad to see that not even among you guys there was a full settlement about this topic. Cheers, jesus - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
On Qua, 2013-01-09 at 12:04 +0200, Thiago Marcos P. Santos wrote: I think the fact that the regular WebKit review process stops at the boundary of WebKit2 should be documented in the WebKit Committers and Reviewer Policy. Agree. And please clarify on the policy if we are talking about everything inside the WebKit2/ directory or if we have exceptions. It is not clear to me if port specific code is covered by this rule and should by reviewed by the owners. And what about code shared by Qt, GTK and EFL (i.e. Platform/CoreIPC/unix/) but not used by Mac? Curious about this myself, I just reviewed a patch only affecting the GTK-specific parts of WebKit2, I believe that is OK? Should we ammend the Owners file to include information about port-specific directories and reviewers? Cheers, -- Gustavo Noronha Silva g...@gnome.org GNOME Project ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
I've got a patch in flight that adds a feature flag. https://bugs.webkit.org/show_bug.cgi?id=106275 According to the instructions liked below I need to edit a WebKit2 file http://trac.webkit.org/wiki/AddingFeatures#ActivatingafeatureforAutotoolsbasedports Does that guideline change? Should I remove the WebKit2 change? On Tue, Jan 8, 2013 at 2:57 PM, Sam Weinig wei...@apple.com wrote: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. 2) Ports must keep themselves building. Non Apple Mac ports, if broken by core functionality changes to WebKit2, are now responsible for fixing themselves. We have asked those who run the EWS bots to make sure that failing to build WebKit2 does not block the commit queue from committing. 3) Over time, owners may remove peripheral functionality from the main WebKit2 directory, such as support for features that aren't broadly applicable. We will not do this immediately, and we will work with ports that are interested in such features to create appropriate, maintainable general-purpose mechanisms that can be used to implement them outside of core WebKit2 code. While we understand that this change will inconvenience some ports, we have decided that forward progress of WebKit2 is a more important concern, and we are moving forward with this change tonight. - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
Trivial changes like this do not need to be approved by an owner. -Sam On Jan 9, 2013, at 9:38 AM, Gregg Tavares g...@google.com wrote: I've got a patch in flight that adds a feature flag. https://bugs.webkit.org/show_bug.cgi?id=106275 According to the instructions liked below I need to edit a WebKit2 file http://trac.webkit.org/wiki/AddingFeatures#ActivatingafeatureforAutotoolsbasedports Does that guideline change? Should I remove the WebKit2 change? On Tue, Jan 8, 2013 at 2:57 PM, Sam Weinig wei...@apple.com wrote: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. 2) Ports must keep themselves building. Non Apple Mac ports, if broken by core functionality changes to WebKit2, are now responsible for fixing themselves. We have asked those who run the EWS bots to make sure that failing to build WebKit2 does not block the commit queue from committing. 3) Over time, owners may remove peripheral functionality from the main WebKit2 directory, such as support for features that aren't broadly applicable. We will not do this immediately, and we will work with ports that are interested in such features to create appropriate, maintainable general-purpose mechanisms that can be used to implement them outside of core WebKit2 code. While we understand that this change will inconvenience some ports, we have decided that forward progress of WebKit2 is a more important concern, and we are moving forward with this change tonight. - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
On Jan 9, 2013, at 8:10 AM, Gustavo Noronha Silva g...@gnome.org wrote: On Qua, 2013-01-09 at 12:04 +0200, Thiago Marcos P. Santos wrote: I think the fact that the regular WebKit review process stops at the boundary of WebKit2 should be documented in the WebKit Committers and Reviewer Policy. Agree. And please clarify on the policy if we are talking about everything inside the WebKit2/ directory or if we have exceptions. It is not clear to me if port specific code is covered by this rule and should by reviewed by the owners. And what about code shared by Qt, GTK and EFL (i.e. Platform/CoreIPC/unix/) but not used by Mac? Curious about this myself, I just reviewed a patch only affecting the GTK-specific parts of WebKit2, I believe that is OK? Should we ammend the Owners file to include information about port-specific directories and reviewers? Cheers, At this point, we ask that all completely non-trivial patches be reviewed by an owner, even if in port specific code. - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
Hi Sam. Some comments below. Cheers, --Antonio Curious about this myself, I just reviewed a patch only affecting the GTK-specific parts of WebKit2, I believe that is OK? Should we ammend the Owners file to include information about port-specific directories and reviewers? Cheers, At this point, we ask that all completely non-trivial patches be reviewed by an owner, even if in port specific code. First, I would like to say that I understand the frustration you guys might have faced by not being able to move Core WebKit2 development at the speed you guys think you could go due to other WebKit2 ports. That is indeed not the goal of the project, and likely the first time the project has seen it at this scale (correct if I am wrong, please). Further, although I do not fully support the direction pointed out as the solution to this problem, I have to agree that it might work. However, I am wondering if the new Core WK2 owners would really feel comfortable in reviewing Qt, Gtk and EFL specific WK2 patches - given that they are likely unfamiliar with the code. Does not it go against the primary *rule* of the WebKit reviewership process, where a reviewer is only allowed to R+ a patch he/she fully understands? If we had a concept of a super-review instead, like Firefox, where owners sometimes rubber stamp patches even when they do not have the know-how to reliably review it, given that the patch has got an r+ of someone else that actually does. Maybe your last email was that one that actually scared me. Looking forward for you reply, ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
[webkit-dev] Changes to the WebKit2 development process
Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. 2) Ports must keep themselves building. Non Apple Mac ports, if broken by core functionality changes to WebKit2, are now responsible for fixing themselves. We have asked those who run the EWS bots to make sure that failing to build WebKit2 does not block the commit queue from committing. 3) Over time, owners may remove peripheral functionality from the main WebKit2 directory, such as support for features that aren't broadly applicable. We will not do this immediately, and we will work with ports that are interested in such features to create appropriate, maintainable general-purpose mechanisms that can be used to implement them outside of core WebKit2 code. While we understand that this change will inconvenience some ports, we have decided that forward progress of WebKit2 is a more important concern, and we are moving forward with this change tonight. - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] Changes to the WebKit2 development process
On Jan 8, 2013, at 2:57 PM, Sam Weinig wei...@apple.com wrote: Hello webkit-dev, We are making some changes to the development process for WebKit2. These changes were announced to reviewers in advance, and I'd like to share them with you now. WebKit2 has a core set of functionality that is valuable to all ports, and then aspects that are only of limited/specialized interest. It is becoming increasingly difficult to improve and advance the core functionality while maintaining the more peripheral aspects. In addition, changes to the core often require significant expertise to evaluate, for instance to ensure that the security and responsiveness goals of WebKit2 are met. The changes are: 1) WebKit2 now has owners. Only owners should review WebKit2 patches. While we do not want to apply this concept across the whole WebKit project at this time, for WebKit2 it is appropriate. The list of owners is documented in the Owners file at the WebKit2 top level directory, and in committers.py. 2) Ports must keep themselves building. Non Apple Mac ports, if broken by core functionality changes to WebKit2, are now responsible for fixing themselves. We have asked those who run the EWS bots to make sure that failing to build WebKit2 does not block the commit queue from committing. I didn't see a settlement on this point of the proposal on previous discussions. Did you elaborate on the feedback that you got when you asked for that the first time? I think it is a question of fair play to not leave core build bots of other platforms broken. That is what we agreed on WebKit and I don't see the reason why it should be different on WebKit2 (which is a part of WebKit). That doesn't mean that the other suggestions aren't reasonable. Greetings, Dirk 3) Over time, owners may remove peripheral functionality from the main WebKit2 directory, such as support for features that aren't broadly applicable. We will not do this immediately, and we will work with ports that are interested in such features to create appropriate, maintainable general-purpose mechanisms that can be used to implement them outside of core WebKit2 code. While we understand that this change will inconvenience some ports, we have decided that forward progress of WebKit2 is a more important concern, and we are moving forward with this change tonight. - Sam ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev ___ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo/webkit-dev
