The deployment branch, originally envisioned on the lists by myself, is intended to provide users of mediawiki with access to the latest stable revision - the one being used on Wikipedia, but not trunk. Unfortunately the deployment branch has been subverted and is now full of WMF specific hacks. Can we please have a the WMF deployment branch be a branch of an untainted branch of the particular revision that is considered stable?
Many thanks! /Brian
On 2009-07-30, Brian wrote:
Can we please have a the WMF deployment branch be a branch of an untainted branch of the particular revision that is considered stable?
This is, and always has been, the latest release (e.g. MediaWiki 1.15.1). If you want to run beyond-stable you cannot expect developers who already don't have enough time to spoon feed you information they don't have. The WMF deployment branch is already well beyond what end users should expect.
On Thu, Jul 30, 2009 at 10:34 AM, Robert Leverington robert@rhl.me.ukwrote:
On 2009-07-30, Brian wrote:
Can we please have a the WMF deployment branch be a branch of an
untainted
branch of the particular revision that is considered stable?
This is, and always has been, the latest release (e.g. MediaWiki 1.15.1). If you want to run beyond-stable you cannot expect developers who already don't have enough time to spoon feed you information they don't have. The WMF deployment branch is already well beyond what end users should expect.
-- Robert Leverington http://rhl.me.uk/
The versioning interactions between mediawiki and its extensions are so complicated that it is simply not feasible to run the latest release while simultaneously using a variety of extensions.
Additionally, I strongly disagree with your definition of the code being used on Wikipedia as beyond stable.
On 2009-07-30, Brian wrote:
The versioning interactions between mediawiki and its extensions are so complicated that it is simply not feasible to run the latest release while simultaneously using a variety of extensions.
Please can you explain more; last time I tried, branched versions of extension generally worked with branched versions of MediaWiki? If this isn't the case then it should be fixed.
Additionally, I strongly disagree with your definition of the code being used on Wikipedia as beyond stable.
Code in trunk is under development, that is what defines trunk - it hasn't been tested to the same level as releases. Although I think this is case where everyone will have different opinions, I'm not sure of the current official stance.
On Thu, Jul 30, 2009 at 10:54 AM, Robert Leverington robert@rhl.me.ukwrote:
On 2009-07-30, Brian wrote:
The versioning interactions between mediawiki and its extensions are so complicated that it is simply not feasible to run the latest release
while
simultaneously using a variety of extensions.
Please can you explain more; last time I tried, branched versions of extension generally worked with branched versions of MediaWiki? If this isn't the case then it should be fixed.
I'm simply referring to the particular versions of extensions that work with particular versions of MediaWiki.
Additionally, I strongly disagree with your definition of the code being used on Wikipedia as beyond stable.
Code in trunk is under development, that is what defines trunk - it hasn't been tested to the same level as releases. Although I think this is case where everyone will have different opinions, I'm not sure of the current official stance.
Trunk has been de facto unstable for a long time and now it really is unstable. For example, when they merged in the upload branch lots of things broke. Now there is a deployment branch which *is* stable. Some may argue of this definition of stability, but it is a de facto stability. The code being run on the 8th largest website in the world is stable. Some of that code isn't needed, or isn't desirable, in the mediawiki installations that the rest of us use.
For what its worth: it has been stated in several places by several people that wmf-deployment is not suitable for end users. It was never intended for use outside the WMF and people depending on it are mistaken.
-Chad
On Jul 30, 2009 12:54 PM, "Robert Leverington" robert@rhl.me.uk wrote:
On 2009-07-30, Brian wrote:
The versioning interactions between mediawiki and its extensions are so >
complicated that it is s... Please can you explain more; last time I tried, branched versions of extension generally worked with branched versions of MediaWiki? If this isn't the case then it should be fixed.
Additionally, I strongly disagree with your definition of the code being >
used on Wikipedia as b... Code in trunk is under development, that is what defines trunk - it hasn't been tested to the same level as releases. Although I think this is case where everyone will have different opinions, I'm not sure of the current official stance.
--
Robert Leverington http://rhl.me.uk/
_______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Thu, Jul 30, 2009 at 11:01 AM, Chad innocentkiller@gmail.com wrote:
For what its worth: it has been stated in several places by several people that wmf-deployment is not suitable for end users. It was never intended for use outside the WMF and people depending on it are mistaken.
-Chad
I'm still waiting on the rational rejection of the need for a deployment branch of mediawiki for the rest of us.
On Thu, Jul 30, 2009 at 11:40 AM, Brian Brian.Mingus@colorado.edu wrote:
On Thu, Jul 30, 2009 at 11:01 AM, Chad innocentkiller@gmail.com wrote:
For what its worth: it has been stated in several places by several people that wmf-deployment is not suitable for end users. It was never intended for use outside the WMF and people depending on it are mistaken.
-Chad
I'm still waiting on the rational rejection of the need for a deployment branch of mediawiki for the rest of us.
I would also like to note that given how trivial it is to create such a branch it is hardly worth a long argument.
On 30/07/2009, at 6:40 PM, Brian wrote:
I'm still waiting on the rational rejection of the need for a deployment branch of mediawiki for the rest of us.
You're welcome to do the work yourself instead of posting a request and expecting somebody to do nontrivial work for you, especially when you complain that it's not being done fast enough.
-- Andrew Garrett agarrett@wikimedia.org http://werdn.us/
On Thu, Jul 30, 2009 at 5:21 PM, Andrew Garrett agarrett@wikimedia.orgwrote:
On 30/07/2009, at 6:40 PM, Brian wrote:
I'm still waiting on the rational rejection of the need for a deployment branch of mediawiki for the rest of us.
You're welcome to do the work yourself instead of posting a request and expecting somebody to do nontrivial work for you, especially when you complain that it's not being done fast enough.
-- Andrew Garrett agarrett@wikimedia.org http://werdn.us/
You seem confused, perhaps you should re-read the thread. I never asked anyone to do anything except make a branch.
On 2009-07-30, Brian wrote:
You seem confused, perhaps you should re-read the thread. I never asked anyone to do anything except make a branch.
That *is* non-trivial. Someone will have to maintain that branch. Someone will have to make a judgement about when to update that branch.
Generally these aren't things that a single person can answer, which is one reason why we have a multiple release canidate policy when declaring versions stable.
On Thu, Jul 30, 2009 at 6:00 PM, Robert Leverington robert@rhl.me.ukwrote:
On 2009-07-30, Brian wrote:
You seem confused, perhaps you should re-read the thread. I never asked anyone to do anything except make a branch.
That *is* non-trivial. Someone will have to maintain that branch. Someone will have to make a judgement about when to update that branch.
Generally these aren't things that a single person can answer, which is one reason why we have a multiple release canidate policy when declaring versions stable.
-- Robert Leverington http://rhl.me.uk/
Fair enough. I was moreso referring to the hoity toity attitude that gasps at the fact that someone might file a bug report rather than fixing it themselves, or makes assumptions about who the bug would be assigned to.
On Thu, Jul 30, 2009 at 6:19 PM, Brian Brian.Mingus@colorado.edu wrote:
The deployment branch, originally envisioned on the lists by myself, is intended to provide users of mediawiki with access to the latest stable revision - the one being used on Wikipedia, but not trunk. Unfortunately the deployment branch has been subverted and is now full of WMF specific hacks. Can we please have a the WMF deployment branch be a branch of an untainted branch of the particular revision that is considered stable?
ِAsking too much?
On Thu, Jul 30, 2009 at 10:45 AM, Mohamed Magdy mohamed.m.k@gmail.comwrote:
On Thu, Jul 30, 2009 at 6:19 PM, Brian Brian.Mingus@colorado.edu wrote:
The deployment branch, originally envisioned on the lists by myself, is intended to provide users of mediawiki with access to the latest stable revision - the one being used on Wikipedia, but not trunk. Unfortunately the deployment branch has been subverted and is now full of WMF specific
hacks.
Can we please have a the WMF deployment branch be a branch of an
untainted
branch of the particular revision that is considered stable?
ِAsking too much?
Until recently there was only trunk. It had been a long standing but unattained goal that the trunk was the live, relatively stable code being run on the site. This was never actually the case, and the deployed code began to fall farther and farther behind trunk. Many people began to complain that there was no reliable way to be sure they were running the same code as on Wikipedia, which is desirable for a variety of reasons. A deployment branch was created. Brion and others are aware that others are running the deployment branch and they did not object, nor advise us not to. But now the deployment branch has foundation specific hacks. Thus, it is desirable that there be another branch available that is marked read only in svn. There are a few ways to implement this, one being a read only branch of the stable revision that is further branched into the wmf-deployment branch. Another implementation is simply to create two branches. I don't think it's asking that much.
On 7/30/09 9:19 AM, Brian wrote:
The deployment branch, originally envisioned on the lists by myself, is intended to provide users of mediawiki with access to the latest stable revision - the one being used on Wikipedia, but not trunk. Unfortunately the deployment branch has been subverted and is now full of WMF specific hacks. Can we please have a the WMF deployment branch be a branch of an untainted branch of the particular revision that is considered stable?
I think you miss the point of the deployment branch, which is to give everyone clean and direct access to and versioning of our actually-deployed software, so we can maintain it easily and other folks can get access to it easily.
It's never been, nor been meant to be, anything else.
**Note that dev trunk and the release branches have WMF-specific hacks too!**
The process of development is such that these things become generalized over time, and then a specific hack is no longer needed. Please feel free to pitch in when you see such things that still need cleanup.
The wmf-deployment shouldn't _need_ any site-specific hacks, just as dev trunk and release versions shouldn't. But the deployment branch, BY DEFINITION, is what we're running in production.
-- brion
On Thu, Jul 30, 2009 at 11:59 AM, Brion Vibber brion@wikimedia.org wrote:
On 7/30/09 9:19 AM, Brian wrote:
The deployment branch, originally envisioned on the lists by myself, is intended to provide users of mediawiki with access to the latest stable revision - the one being used on Wikipedia, but not trunk. Unfortunately
the
deployment branch has been subverted and is now full of WMF specific
hacks.
Can we please have a the WMF deployment branch be a branch of an
untainted
branch of the particular revision that is considered stable?
I think you miss the point of the deployment branch, which is to give everyone clean and direct access to and versioning of our actually-deployed software, so we can maintain it easily and other folks can get access to it easily.
It's never been, nor been meant to be, anything else.
**Note that dev trunk and the release branches have WMF-specific hacks too!**
The process of development is such that these things become generalized over time, and then a specific hack is no longer needed. Please feel free to pitch in when you see such things that still need cleanup.
The wmf-deployment shouldn't _need_ any site-specific hacks, just as dev trunk and release versions shouldn't. But the deployment branch, BY DEFINITION, is what we're running in production.
-- brion
Thanks for the clarification. I will file bugs for these issues in the future, then.
On 7/30/09 11:02 AM, Brian wrote:
The process of development is such that these things become generalized over time, and then a specific hack is no longer needed. Please feel free to pitch in when you see such things that still need cleanup.
The wmf-deployment shouldn't _need_ any site-specific hacks, just as dev trunk and release versions shouldn't. But the deployment branch, BY DEFINITION, is what we're running in production.
Thanks for the clarification. I will file bugs for these issues in the future, then.
Awesome, thanks! :D
My ideal is certainly that wmf-deployment should just be a delayed mirror of trunk 99% of the time.
-- brion
wikitech-l@lists.wikimedia.org