+1 - YES please. Requiring cordoba-common for my react-native-cordova-plugin 
adapter was a nightmare !! 




On 10/20/15, 2:23 PM, "Nikhil Khandelwal" <nikhi...@microsoft.com> wrote:

>+1 to publishing cordova-common to npm. 
>
>-Nikhil
>
>-----Original Message-----
>From: Steven Gill [mailto:stevengil...@gmail.com] 
>Sent: Tuesday, October 20, 2015 2:20 PM
>To: dev@cordova.apache.org
>Subject: Re: [Discuss] Cordova-common release
>
>I want to revisit this.
>
>So cordova-android has a dependency now on cordova-common. It is a 
>bundledDependency so when we generate a tar to release cordova-android, it 
>will be included. It will also be in the cordova-android package that gets 
>downloaded with cordova platform add.
>
>This is fine for released work, but more annoying for development. I need to 
>npm link cordova-common into cordova-android (and soon every platform which 
>implements common platformAPI). We could check in cordova-common into 
>cordova-android but that isn't a great solution either.
>
>I agree that we should be going towards smaller modules and not having a case 
>of cordovaLib1, cordovaLib2, etc. I think this is still going to be a work in 
>progress and will take some time.
>
>For the interim, I recommend we publish cordova-common. Of course, continue to 
>add it as a bundledDependency so users don't need to npm install it with 
>released packages.
>
>On Wed, Sep 30, 2015 at 7:24 AM, Vladimir Kotikov (Akvelon) < 
>v-vlk...@microsoft.com> wrote:
>
>> > I still do not understand what are you trying to solve by having all
>> that content published as big blob.
>> Code deduplication is the main reason. All the things from 
>> 'cordova-common' will be used by platforms intensively, so we need to 
>> share this code and keep it separately from LIB to share easily. 
>> Publishing is basically doesn't required for this, and bundling 
>> 'cordova-common' into LIB is enough for this purpose.
>>
>> Another reason was that third-party tool might want to use some of 
>> this functionality (like your example with ConfigParser), so we need 
>> to have this package on NPM to allow them to get it. For this case I 
>> now do agree with you that separate packages for ConfigParser, 
>> PluginInfo and other stuff looks better than putting it into one big package.
>>
>> -
>> Best regards, Vladimir
>>
>>
>> -----Original Message-----
>> From: Carlos Santana [mailto:csantan...@gmail.com]
>> Sent: Wednesday, September 30, 2015 2:07 PM
>> To: dev@cordova.apache.org
>> Subject: Re: [Discuss] Cordova-common release
>>
>> Yes temporary, maybe we can discuss some more in F2F
>>
>> I still do not understand what are you trying to solve by having all 
>> that content published as big blob.
>>
>> If the packages are only for Cordova components to depend on then we 
>> control the release and we can include them easily.
>>
>> If the code is to be share by third party or anyone out there then it 
>> make sense to put in npm.
>>
>> One concrete example is cordova-configparser, Our IBM tool is using it 
>> in our own models code so today we taking a copy, if it's available 
>> thru npm then we can stated as a dependency and manage it as a npm 
>> package vs a loosely node module js file
>>
>> Maybe not all classes need to be converted to npm packages maybe it 
>> can be some cordova-configparser cordova-utils cordova-helper
>>
>> Also do some refactoring and dependency cleaning, I saw a node module 
>> dependeding on underscore and the file only had one simple call to 
>> _.find()
>>
>> We were going to use that module, but then decided not to since it 
>> depended on underscore for a simple thing, this creates legal 
>> clearance work and more dependencies we need to include in our product 
>> making our download larger.
>>
>> And yes, yes we bundle all our dependencies when we go into production.
>>
>> - Carlos
>> Sent from my iPhone
>>
>> > On Sep 30, 2015, at 5:27 AM, Vladimir Kotikov (Akvelon) <
>> v-vlk...@microsoft.com> wrote:
>> >
>> > Including cordova-common as bundled dependency should be enough in 
>> > our
>> case. Changes to coho are submitted already [1], so we only need to 
>> update package.json for cordova-lib.
>> >
>> > Personally  for me bundling looks like a temporary workaround before 
>> > we
>> get all those partials of 'common' published to npm, am I right?
>> >
>> > [1]
>> > https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fgit
>> > hu
>> > b.com%2fapache%2fcordova-coho%2fpull%2f94&data=01%7c01%7cv-vlkoti%40
>> > 06
>> > 4d.mgd.microsoft.com%7cf9eefe800e4c44c0540308d2c9874d65%7c72f988bf86
>> > f1 
>> > 41af91ab2d7cd011db47%7c1&sdata=HpV5fWkx6nUZUU%2fT4qVVo0QZiGM7%2fm%2b
>> > do
>> > 9WX4V4JfT0%3d
>> >
>> > -
>> > Best regards, Vladimir.
>> >
>> > -----Original Message-----
>> > From: Carlos Santana [mailto:csantan...@gmail.com]
>> > Sent: Tuesday, September 29, 2015 9:15 PM
>> > To: dev@cordova.apache.org
>> > Subject: Re: [Discuss] Cordova-common release
>> >
>> > Do we need to absolutely publish this to npm?
>> >
>> > Can we just include the dependency in the platform as a bundle
>> dependency?
>> >
>> > We just need to update coho to npm install/link the cordoba-common 
>> > package when doing a release of what ever component need it? (i.e.
>> > cordova-android)
>> >
>> > Will this get you what you want? Why does it absolutely need to be 
>> > in
>> npm registry?
>> >
>> > I really don't think will be a good idea to publish two npm packages
>> "cordova-lib" and "cordova-common"
>> >
>> > Sorry if I'm being a pain in the ass, maybe I'm something obvious 
>> > here
>> >
>> >
>> >> On Tue, Sep 29, 2015 at 1:34 PM Steven Gill 
>> >> <stevengil...@gmail.com>
>> wrote:
>> >>
>> >> Sounds good. Let's move forward
>> >> On Sep 29, 2015 10:21 AM, "Nikhil Khandelwal"
>> >> <nikhi...@microsoft.com>
>> >> wrote:
>> >>
>> >>> +1. I understand the value of Carlos' proposal, but in the spirit 
>> >>> +of
>> >>> moving forward with this which is fairly complicated refactor 
>> >>> involving multiple releases and repos, I would like us to make 
>> >>> progress on this
>> >> soon
>> >>> and not add significant scope to this effort.
>> >>>
>> >>>
>> >>> -Nikhil
>> >>>
>> >>> -----Original Message-----
>> >>> From: Sergey Grebnov (Akvelon) [mailto:v-seg...@microsoft.com]
>> >>> Sent: Tuesday, September 29, 2015 1:34 AM
>> >>> To: dev@cordova.apache.org
>> >>> Subject: RE: [Discuss] Cordova-common release
>> >>>
>> >>> +1
>> >>>
>> >>> -----Original Message-----
>> >>> From: Vladimir Kotikov (Akvelon) [mailto:v-vlk...@microsoft.com]
>> >>> Sent: Tuesday, September 29, 2015 11:27 AM
>> >>> To: dev@cordova.apache.org
>> >>> Subject: RE: [Discuss] Cordova-common release
>> >>>
>> >>> Agree with you, guys.
>> >>>
>> >>> Unfortunately, the underlying modules in `cordova-common` are not 
>> >>> really atomic, since they depending on each other. For example 
>> >>> ConfigParser requires `xmlHelpers`, `events` and `CordovaError` as 
>> >>> a
>> dependencies.
>> >>> Reworking them to be truly separated might be sort of problematic, 
>> >>> especially in context of message logging (as they use shared event
>> >> emitter
>> >>> to log output to console).
>> >>>
>> >>> So I still propose is to release `common` module as-is and then 
>> >>> gradually move inner modules out to separate packages.
>> >>>
>> >>> -
>> >>> Best regards, Vladimir.
>> >>>
>> >>> -----Original Message-----
>> >>> From: Carlos Santana [mailto:csantan...@gmail.com]
>> >>> Sent: Friday, September 25, 2015 7:33 PM
>> >>> To: dev@cordova.apache.org
>> >>> Subject: Re: [Discuss] Cordova-common release
>> >>>
>> >>> Sorry a typo
>> >>> to use "bundleDependencies" you will have a node_modules/ 
>> >>> directory directly under "common/node_modules/cordova-error/"
>> >>>
>> >>> and the the small modules (i.e. cordoba-util, cordova-plugin-info,
>> >>> etc..) will be located there.
>> >>>
>> >>> then have explicit ignores for the dependencies you don't want to 
>> >>> be source control like npm [2]
>> >>>
>> >>> [2]:
>> >> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fgi
>> >> th
>> >> u
>> >> b.com%2fnpm%2fnpm%2fblob%2fmaster%2f.gitignore%23L24&data=01%7c01%7
>> >> cv
>> >> -
>> >> vlkoti%40064d.mgd.microsoft.com%7c73b4ff38f0fe41e1f18608d2c5c70e0f%
>> >> 7c
>> >> 7
>> >> 2f988bf86f141af91ab2d7cd011db47%7c1&sdata=tU%2bFHDUJZXzXnbG%2fUP7AY
>> >> 4q
>> >> E
>> >> CnvsbnsJ%2bvEriJvqYcU%3d
>> >>>
>> >>>
>> >>> On Fri, Sep 25, 2015 at 12:24 PM Carlos Santana 
>> >>> <csantan...@gmail.com>
>> >>> wrote:
>> >>>
>> >>>> Yes after reviewing the changes, I understood the purpose of the 
>> >>>> code that you seperated to avoid duplicate code between the other 
>> >>>> top level modules (i.e. platforms, lib, cli)
>> >>>>
>> >>>> I still think small modules is the way to go.
>> >>>>
>> >>>> Don't let process, bureaucracy, and ceremony hamper the 
>> >>>> engineering process and the consumer UX using this modules.  
>> >>>> (yeah that came out from the mouth of an IBMer ;-p)
>> >>>>
>> >>>> Yes, I'm not blind, I understand the voting, the releasing, the 
>> >>>> packaging the publish steps
>> >>>>
>> >>>> The way I look at it no matter what you do you are not going to 
>> >>>> make it easier by having one "common" package.
>> >>>>
>> >>>> Let's say you just need to update a file to fix a bug from Error, 
>> >>>> well you need to test, vote, release, "common"
>> >>>> Next you want to fix a bug in configparser, guess what you need 
>> >>>> to tests, vote, release "common" again But if only config parser 
>> >>>> changed all the rest of the code in "common"
>> >>>> needs to be tested and release, and consumer will need to pick a 
>> >>>> new common for only a small bug fix in a portion of "common"
>> >>>>
>> >>>> Basically that's what we have today, the way I see it you are 
>> >>>> just creating two libs "lib" and "lib2"
>> >>>>
>> >>>> With large number of small modules, lets say we create 8 now, 
>> >>>> maybe only 2 change most of the time, and the other 5 are so 
>> >>>> basic and small that they never change and stay at version 1.0.0 for  
>> >>>> long time.
>> >>>>
>> >>>> I think for this small modules, I don't think we have to do the 
>> >>>> blog post, twitter things, those I will continue to have for the 
>> >>>> large components (cli, platforms, plugins)
>> >>>>
>> >>>> Also the side effect I would like to see, is clean APIs edges, 
>> >>>> each small module provides an API, it contain tests to that API, 
>> >>>> and lib contain integration tests as a whole.
>> >>>>
>> >>>> Maybe the compromise for now, to move forward let's break the 
>> >>>> functions into "npm packages" inside this "common" where each sub 
>> >>>> directory inside common is a npm package with a single entry 
>> >>>> point
>> >>>> (index.js) and common package.json have them as 
>> >>>> "bundleDependencies", similar way as npm does [1]
>> >>>>
>> >>>> the transition will be for consumers for their dependencies and 
>> >>>> the way they consume the API
>> >>>> dependencies: {
>> >>>>   cordova-common: "1.0.0"
>> >>>> }
>> >>>> cordova-common only expose one index.js with the APIs to the 
>> >>>> other modules
>> >>>>
>> >>>> var cdvUtil              = require("cordova-common").cordova-util
>> >>>> cdvPluginInfo          =
>> require("cordova-common").cordova-plugin-info,
>> >>>> cdvError                  = require("cordova-common").cordova-error,
>> >>>> cdvConfigParser     = require("cordova-common").cordova-config-parser,
>> >>>> cdvConfigChanges =
>> >>>> require("cordova-common").rcordova-config-changes);
>> >>>>
>> >>>> then it can be easier transition if we want to change later, 
>> >>>> nothing much on our part since we already have the npm packages 
>> >>>> implemented it's a matter if we want to make them available on 
>> >>>> npm or
>> not.
>> >>>>
>> >>>> [1]:
>> >>>> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2f
>> >>>> g
>> >>>> ithu
>> >>>> b.com%2fnpm%2fnpm%2fblob%2fmaster%2fpackage.json%23L137&data=01%7
>> >>>> c
>> >>>> 01%7
>> >>>> cv-vlkoti%40064d.mgd.microsoft.com%7c73b4ff38f0fe41e1f18608d2c5c7
>> >>>> 0
>> >>>> e0f%
>> >>>> 7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=tUdBPvbE8oC3DPubmVx5
>> >>>> 0
>> >>>> QKD2
>> >>>> CLfoxrVgj%2ftTxTrMJ8%3d
>> >>>>
>> >>>>
>> >>>>
>> >>>>
>> >>>>
>> >>>> On Fri, Sep 25, 2015 at 11:40 AM Sergey Grebnov (Akvelon) < 
>> >>>> v-seg...@microsoft.com> wrote:
>> >>>>
>> >>>>> I tend to agree w/ Carlos here, but from practical side it might 
>> >>>>> be very hard to maintain and release such a granular modules, 
>> >>>>> for example,
>> >>>>> -  cordova-error has been updated ->Vote -> update 
>> >>>>> cordova-config-parser
>> >>>>> + Vote-> update + Vote other depended modules
>> >>>>> - now we want to add some new feature: modules are very granular 
>> >>>>> so we should introduce a new module
>> >>>>>
>> >>>>> But I totally love and support Carlos idea regarding grouping 
>> >>>>> meaningful/independent logic in modules, this is how software 
>> >>>>> must be designed.
>> >>>>>
>> >>>>> I personally think about this new module as some sort of core 
>> >>>>> Cordova functionality and high level classes which could be used 
>> >>>>> by cordova-lib/cli and platforms -unified CordovaError, events 
>> >>>>> (output tracing, etc), working with config file, superspawn, etc
>> >>>>>
>> >>>>> Thx!
>> >>>>> Sergey
>> >>>>> -----Original Message-----
>> >>>>> From: Carlos Santana [mailto:csantan...@gmail.com]
>> >>>>> Sent: Thursday, September 24, 2015 6:31 PM
>> >>>>> To: dev@cordova.apache.org
>> >>>>> Subject: Re: [Discuss] Cordova-common release
>> >>>>>
>> >>>>> Sorry if I take my inner purist theoretical developer out for a 
>> >>>>> minute :-)
>> >>>>>
>> >>>>> The point of having a "node module" it should be explicit and 
>> >>>>> small, meaning that should be easy to name in a way that 
>> >>>>> describes what it is or do.
>> >>>>>
>> >>>>> Take into account that "node module" is not the same as a "npm
>> >> package"
>> >>>>>
>> >>>>> Having 2 npm packages on the npm registry "cordova-common" and 
>> >>>>> "cordova-lib" to the simple eye would look like duplicate 
>> >>>>> packages, and then will need to answer multiple times "What is 
>> >>>>> the difference between lib and common?"
>> >>>>>
>> >>>>> Why not have more smaller explicit npm packages instead?
>> >>>>>
>> >>>>> cordova-util
>> >>>>> cordova-plugin-info
>> >>>>> cordova-error
>> >>>>> cordova-config-parser
>> >>>>> cordova-config-changes
>> >>>>>
>> >>>>> each one with a index.js exposing APIs
>> >>>>>
>> >>>>> Then the programing model becomes something like this:
>> >>>>> var cdvUtil              = require('cordova-util'),
>> >>>>> cdvPluginInfo          = require('cordova-plugin-info'),
>> >>>>> cdvError                  = require('cordova-error'),
>> >>>>> cdvConfigParser     = require('cordova-config-parser'),
>> >>>>> cdvConfigChanges = require('cordova-config-changes');
>> >>>>>
>> >>>>>
>> >>>>> On Thu, Sep 24, 2015 at 12:29 PM Sergey Grebnov (Akvelon) < 
>> >>>>> v-seg...@microsoft.com> wrote:
>> >>>>>
>> >>>>>> Hi guys - we've decided to combine shared logic as 
>> >>>>>> cordova-common module and publish it separately (read [1] for more 
>> >>>>>> details).
>> >>>>>> Corresponding change has landed to master[2] on last week so 
>> >>>>>> I'm wondering if we should release this module and then update 
>> >>>>>> LIB to rely
>> >>>>> on it (similar to cordova-serve).
>> >>>>>> So guys it will be great if we can review it one more time 
>> >>>>>> (including the name - do we all  agree to use cordova-common??) 
>> >>>>>> and then do release - I'll be able to help w/ merging the 
>> >>>>>> recent changes added to LIB before doing release.
>> >>>>>>
>> >>>>>> [1]
>> >>>>>> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%
>> >>>>>> 2fis
>> >>>>>> sue
>> >>>>>> s.apache.org%2fjira%2fbrowse%2fCB-9598&data=01%7c01%7cv-segreb%
>> >>>>>> 40mi
>> >>>>>> cro
>> >>>>>> soft.com%7cf31529ebb0de4bf28ebd08d2c54915f3%7c72f988bf86f141af9
>> >>>>>> 1ab2
>> >>>>>> d7c
>> >>>>>> d011db47%7c1&sdata=oeX8CbX%2bQGJsvf9%2fW2KFWAkUw6NAlb0LMOorTjwX
>> >>>>>> TXk%
>> >>>>>> 3d
>> >>>>>> [2]
>> >>>>>> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%
>> >>>>>> 2fgi
>> >>>>>> thu
>> >>>>>> b.com%2fapache%2fcordova-lib%2ftree%2fmaster%2fcordova-common&d
>> >>>>>> ata=
>> >>>>>> 01%
>> >>>>>> 7c01%7cv-segreb%40microsoft.com%7cf31529ebb0de4bf28ebd08d2c5491
>> >>>>>> 5f3%
>> >>>>>> 7c7
>> >>>>>> 2f988bf86f141af91ab2d7cd011db47%7c1&sdata=o0EwRydALocXUrr4I9ASf
>> >>>>>> QMku
>> >>>>>> RV0
>> >>>>>> ksMKA%2fp2zpg6BNU%3d
>> >>>>>>
>> >>>>>> Thx!
>> >>>>>> Sergey
>> >>>>>>
>> >>>>>> ---------------------------------------------------------------
>> >>>>>> ----
>> >>>>>> -- To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
>> >>>>>> For additional commands, e-mail: dev-h...@cordova.apache.org
>> >>>
>> >>> ------------------------------------------------------------------
>> >>> --
>> >>> - To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
>> >>> For additional commands, e-mail: dev-h...@cordova.apache.org
>> >
>> > --------------------------------------------------------------------
>> > - To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
>> > For additional commands, e-mail: dev-h...@cordova.apache.org
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
>> For additional commands, e-mail: dev-h...@cordova.apache.org
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
>> For additional commands, e-mail: dev-h...@cordova.apache.org
>>
>>
>?B�KKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKCB�?�?[��X��ܚX�K??K[XZ[?�??]�][��X��ܚX�P?�ܙ?ݘK�\?X�?K�ܙ�B��܈?Y??]?[ۘ[??��[X[�?�??K[XZ[?�??]�Z?[???�ܙ?ݘK�\?X�?K�ܙ�B

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@cordova.apache.org
For additional commands, e-mail: dev-h...@cordova.apache.org

Reply via email to