Advice + proposals regarding automodule naming

[forax at univ-mlv.fr]

Hi Brian, 

> De: "Brian Fox" <brianf at sonatype.com>
> À: forax at univ-mlv.fr
> Cc: "Robert Scholte" <rfscholte at apache.org>, jpms-spec-experts at openjdk.java.net
> Envoyé: Jeudi 19 Janvier 2017 15:09:17
> Objet: Re: Advice + proposals regarding automodule naming

> Hi Rémi,

> This isn't a maven problem...it's potentially a problem for _everyone_.

potentially if build tools do nothing. 

> The only reason Maven is introduced into the conversation is because for all the
> stuff in Central, it is a source of a sensible default for a module name,
> otherwise it's completely unrelated.

I agree that Maven Central has sensible defaults so these values can be injected into jars that are not modular. 
The other solution, which do not requires any information is to create the module-info directly from the bytecodes with the caveat that dynamic dependencies (Class.forName, etc) will not be visible. 

The later solution can not be done by the VM because it's slow, a build tool can do that once while the VM will have to do that everytime the application is started. 

Rémi 

> On Thu, Jan 19, 2017 at 9:07 AM, < forax at univ-mlv.fr > wrote:

>> Hi Brian,
>> Maven can decide to put modular jars (a jar that contains a module-info.class)
>> in the modulepath and plain old jars in the classpath,
>> it will work with all existing applications and because it does not use the
>> automatic module feature so there is no naming issue.

>> The problem is that this solution do not offer a clean way to upgrade a jar from
>> the classpath to the modulepath because this solution requires all dependencies
>> to be modular first.

>> Rémi

>>> De: "Brian Fox" < brianf at sonatype.com >
>>> À: forax at univ-mlv.fr
>>> Cc: "Robert Scholte" < rfscholte at apache.org >,
>>> jpms-spec-experts at openjdk.java.net
>>> Envoyé: Jeudi 19 Janvier 2017 14:54:32

>>> Objet: Re: Advice + proposals regarding automodule naming

>>> Not sure if this will get through Robert....
>>> We seem to have diverted away from the main issue. The biggest and most urgent
>>> issue is not how Maven will/won't map directly to modules in the future. It's
>>> an issue to be sure, but the universe of all previously developed Java
>>> components are subject to the auto module behavior and all the issues laid out
>>> in the original mail. If we don't get that fixed in the beginning, it will be
>>> very difficult to change later. Reference the NPM scope issue I cited
>>> originally.

>>> On Tue, Jan 17, 2017 at 5:11 PM, < forax at univ-mlv.fr > wrote:

>>>> Robert,
>>>> i fully agree with you that Maven can not use automatic modules.
>>>> Automatic modules have weird name rules, everything is exported and has no
>>>> dependency itself*, so they are useless if you already have already a trove of
>>>> info like the Maven POM.

>>>> In my opinion, the real question is not how to map existing Maven artifacts to
>>>> Java modules but more,
>>>> how Maven 4 artifacts are mapped to Java modules and then how to make the
>>>> transition between Maven 3 artifacts to Maven 4 artifacts as smooth as
>>>> possible.

>>>> Here is my take on what can be a Maven 4 artifact,
>>>> - a Maven 4 artifact can only depends other Maven 4 artifact (and their are some
>>>> way to see a Maven3 artifact as a Maven 4 artifact if the POM is siple enough),
>>>> - a Maven 4 artifact do not allow split packages (a lot of Maven 3 artifact uses
>>>> split packages because it's a cool way to do an after the fact modularisation
>>>> without changing the name of the module)
>>>> - a Maven 4 artifact info is specified with info extracted from the module-info
>>>> and from the POM
>>>> (version is in the POM, exported packages are in the module-info, ...)
>>>> etc.

>>>> once you have the precise rules, it will be easier to see how to map a Maven 3
>>>> artifact to a Maven 4 and what are the compatibility rules.

>>>> regards,
>>>> Rémi

>>>> * apart if you want to play with configurations that mix modulepath and
>>>> classpath but these kind of configurations are really hard to debug.

>>>> ----- Mail original -----
>>>> > De: "Robert Scholte" < rfscholte at apache.org >
>>>> > À: "Remi Forax" < forax at univ-mlv.fr >
>>>> > Cc: jpms-spec-experts at openjdk.java.net , "Brian Fox" < brianf at sonatype.com >
>>>> > Envoyé: Mardi 17 Janvier 2017 13:04:08
>>>> > Objet: Re: Advice + proposals regarding automodule naming

>>>> > Hi Rémi,

>>>> > In the end every non-jdk.* and non-java.* module in the module-info will
>>>> > be a dependency in your buildtool descriptor. Such module must match
>>>> > exactly one versionless dependency, or conflictId as we call it, which is
>>>> > in general the groupId + artifactId (type and classifier are not relevant
>>>> > for this story).
>>>> > By ignoring the groupId a module can referred by multiple dependencies. So
>>>> > we can expect collissions. For that reason Brian did a quick scan over
>>>> > Maven Central to count the number of duplicate artifactIds.

>>>> > Here's the artifactIds with 100+ groupIds:
>>>> > maven_artifact_id count(DISTINCT maven_group_id) count(maven_group_id)
>>>> > library 391 6854
>>>> > core 312 8188
>>>> > common 142 5084
>>>> > ui 138 1414

>>>> > In theory I could have a Maven project with 391 'library'-jars on the
>>>> > classpath without any problem. And as long as they are direct dependencies
>>>> > I have control over this by simply not adding 'library' as requirement to
>>>> > module-info. The issues start when different 'library'-jars are transitive
>>>> > dependencies and when they are marked are required in the module-info file
>>>> > of my direct or transitive dependencies.

>>>> > Developers of the 'library'-jars cannot use library as the module name and
>>>> > are forced to pick another name. As developer of my project in the end I
>>>> > decide which versions of dependencies are used. If the 'library'-jar gets
>>>> > a different module name and my dependency is still referring to the old
>>>> > module name, the project can't be built.

>>>> > What I expect is that developers are forced to remove the requirements
>>>> > from their module-info because of the mentioned issues. So instead of
>>>> > increasing the number requirements it will be reduced. For that reason we
>>>> > say either use a unique module name from the beginning (GA) or wait until
>>>> > a dependency has its own module name before adding it as requirement.

>>>> > As far as I know this is the first time the JDK/JRE decides (proposes) a
>>>> > name for an entity based on another entity. There are no relations between
>>>> > method-, class-, or package-names and there doesn't have to be a relation
>>>> > between the module name and the filename, so please don't try to do so.

>>>> > regards,
>>>> > Robert

>>>> > On Mon, 16 Jan 2017 16:44:03 +0100, Remi Forax < forax at univ-mlv.fr > wrote:
>>>> >> Hi Robert,
>>>> >> the problem with automatic modules is more general that just the name,
>>>> >> automatics modules also creates a flat hierarchy which doesn't map well
>>>> >> with the Maven artifact descriptor.

>>>> >> I wonder why you want Maven to use automatic modules, or said
>>>> >> differently Maven has a lot of information about the artifact, why do
>>>> >> you want to forget all these information when fetching a Maven artifact.

>>>> >> I think that one problem is that you do not want to create a
>>>> >> module-info.class from the Maven POM and insert it into the jar because
>>>> >> it will change the artifact*.
>>>> >> This kind of modules is supported by jigsaw under the name of synthetic
>>>> >> modules. A synthetic module is a module with a module descriptor not
>>>> >> created by javac but by another tool.

>>>> >> In my opinion, automatic modules are interesting when you have jar that
>>>> >> do not come from Maven central but comes from an ad-hoc build tool and
>>>> >> will be considered as a leaf of the dependency DAG.
>>>> >> Otherwise, for existing module system, using a synthetic module seem to
>>>> >> be a better idea.

>>>> >> regards,
>>>> >> Rémi

>>>> >> * given you have also the problem of split packages, you also need a way
>>>> >> to merge several artifacts into one modular jar because it's the easy
>>>> >> way to solve the split package problem.

>>>> >> ----- Mail original -----
>>>> >>> De: "Robert Scholte" < rfscholte at apache.org >
>>>> >>> À: jpms-spec-experts at openjdk.java.net
>>>> >>> Cc: "Apache Maven Dev" < dev at maven.apache.org >
>>>> >>> Envoyé: Lundi 16 Janvier 2017 10:37:08
>>>> >>> Objet: Advice + proposals regarding automodule naming

>>>> >>> This is a message from Robert Scholte and Brian Fox. We both have been
>>>> >>> talking about this topic several weeks with other Maven developers and
>>>> >>> came to the conclusion that we should warn the jigsaw team with their
>>>> >>> current approach regarding auto modules. We will share our experiences,
>>>> >>> thoughts, conclusions and will suggest two proposals.

>>>> >>> Traditionally, the Java ecosystem has been very mature in terms of
>>>> >>> naming
>>>> >>> and namespacing. The reverse fqdn introduced into the java package was a
>>>> >>> great choice to ensure classes don’t conflict. Popular build tools such
>>>> >>> as
>>>> >>> Maven and nearly all those that followed built upon that this key
>>>> >>> concept
>>>> >>> with the introduction of “GroupId” also using the fqdn as part of the
>>>> >>> name
>>>> >>> to ensure the coordinates were properly namespaced.

>>>> >>> We’ve seen some ecosystems diverge from this leading to new challenges
>>>> >>> that ultimately had to be reversed. A great example can be seen in the “
>>>> >>> tragic mistake from npm creators ” [1] which was to launch without a
>>>> >>> namespace concept. Eventually, NPM started running out of useful names
>>>> >>> and
>>>> >>> had to backtrack to introduce “scopes” which is really just a namespace
>>>> >>> [2]. The real problem here is that the major change in namespace was
>>>> >>> backed in after several years of momentum without it. It’s taken a long
>>>> >>> time for tooling and best practice to catch up to scopes and in the
>>>> >>> interim, people have been left with a dual mode, some namespaced, some
>>>> >>> not
>>>> >>> namespaced situation that has created chaos. [3]

>>>> >>> The real issue at hand here as we consider behaviors in the jigsaw
>>>> >>> automodule revolves around two well studied concepts.

>>>> >>> The most important is the “Default effect” [3] which states that
>>>> >>> whatever
>>>> >>> the default behavior is will become the most prominent best practice. A
>>>> >>> default that uses a filename to generate a very short, un-namespaced
>>>> >>> module id effectively sets the behavior to create generic names that
>>>> >>> will
>>>> >>> eventually conflict...exactly what we’ve seen in npm.

>>>> >>> Additionally, The switching costs introduced in overcoming a default
>>>> >>> un-namespaced module id to one with a unique namespace is also
>>>> >>> significant
>>>> >>> once you consider all the potential users. This is why API change is
>>>> >>> hard,
>>>> >>> and changing the module id after the fact from the default is
>>>> >>> effectively
>>>> >>> an API change.

>>>> >>> The second principal at hand is the “Principle of least astonishment”.
>>>> >>> We
>>>> >>> want to find a default that doesn’t violate what most users would
>>>> >>> consider
>>>> >>> to be the most obvious. One could argue the current auto module
>>>> >>> algorithm
>>>> >>> doesn’t violate this principle, but it’s important to consider alternate
>>>> >>> suggestions in this light.

>>>> >>> First, lets explore the potential downsides if the default effect takes
>>>> >>> hold with the currently generated auto module id. In Apache Maven, the
>>>> >>> artifact id is the part of the coordinate that generates the filename.
>>>> >>> This means that com.somecompany:artifact:version will become
>>>> >>> artifact-version.jar, which would result in automodule id “artifact”.
>>>> >>> Armed with this understanding, that does an analysis of the Maven
>>>> >>> ecosystem have to say about potential conflicts in the automodule id?

>>>> >>> If we ignore the groupid and version of all the components in the Maven
>>>> >>> Central repository, we end up with over 13,500 (7% of the total
>>>> >>> group:artifact combinations) conflicts. This does not consider conflicts
>>>> >>> across other repositories, or within customer portfolios yet it is
>>>> >>> pretty
>>>> >>> telling. Conflicts will happen. In some cases, the number of conflicts
>>>> >>> on
>>>> >>> the same common names is well above 100. The list of conflicts as of
>>>> >>> October, 2016 can be seen here. [6]

>>>> >>> At this point, hopefully we’ve made the case for at least establishing a
>>>> >>> default module id that
>>>> >>> 1. Uses namespaces to minimizes id conflicts when possible
>>>> >>> 2. Leverages the default effect to create a de facto best practice
>>>> >>> 3. Follows the principle of least astonishment

>>>> >>> We have two potential proposals that solve these goals.

>>>> >>> Proposal 1: Leverage existing coordinates when available.

>>>> >>> Maven is inarguably the most popular build system for Java components,
>>>> >>> with Maven Central being the default and largest repository of Java
>>>> >>> components in the world. By default, every jar built by Maven
>>>> >>> automatically gets a simple properties file inserted into it with its
>>>> >>> unique coordinates. Now, not every jar in Central was built with Maven,
>>>> >>> however 94% of them were, as we can find the pom.properties file in
>>>> >>> 1,806,023 of the 1,913,561 central components . Talk about the default
>>>> >>> effect in action!

>>>> >>> It’s further important to recognize that given a jar with a
>>>> >>> pom.properties
>>>> >>> declaring coordinates, it means that the project itself has chosen those
>>>> >>> coordinates as their own name. In other words, this is how they refer to
>>>> >>> themselves, even if other consumers may not be using Maven directly.

>>>> >>> If automodule were able to peek inside a jar and generate the default id
>>>> >>> using the groupid and artifactid present in the file, this would nearly
>>>> >>> eliminate all instances of id conflict because a significant portion of
>>>> >>> the Java ecosystem is in fact built with Maven. Additionally, the fact
>>>> >>> that 1.8 million (and counting) modules would have namespace as the
>>>> >>> default behavior means we’ve taken a huge step in setting the best
>>>> >>> practice of picking module ids with a namepace. Additionally, since the
>>>> >>> project itself has chosen these coordinates and uses them as their
>>>> >>> primary
>>>> >>> distribution mechanism, this follows the principle of least astonishment
>>>> >>> to consumers regardless of their chosen build system. Finally, since all
>>>> >>> of the above are true, it’s unlikely the project would need to migrate
>>>> >>> to
>>>> >>> a new module id when they adopt jigsaw natively, thus avoiding an API
>>>> >>> switching cost for their users.

>>>> >>> Proposal 2: Drop automodules
>>>> >>> Right now Jigsaw tries to calculate a module name solely based on the
>>>> >>> name
>>>> >>> of the jar file, which now already causes issues. Besides the fact that
>>>> >>> the module name is not guaranteed unique compared with its Maven
>>>> >>> coordinate, there are extra transformations which makes it even less
>>>> >>> guaranteed that it is unique; e.g. dashes are replaced by dots (which
>>>> >>> are
>>>> >>> both valid artifactId characters), in some cases the number and their
>>>> >>> following characters are stripped off. For artifacts like
>>>> >>> jboss-servlet-api_4.0_spec it makes sense, however we already see issues
>>>> >>> here where commons-lang, commons-lang2 and commons-lang3 get the same
>>>> >>> module name,
>>>> >>> even though they have different artifactIds and contain different
>>>> >>> packages. Choosing different artifactIds and packages was a very wise
>>>> >>> decision because it made it possible that these jars could live next to
>>>> >>> each other. Removing that separation by the authors is a very unwise
>>>> >>> decision.

>>>> >>> Another known example is the jsrNNN jars, which now all get jsr as the
>>>> >>> module name.

>>>> >>> Is it highly unlikely there is one single rule to capture all the use
>>>> >>> cases and which always result in a module name we can work with.

>>>> >>> For that reason the other proposal is to simply drop automodules. Don’t
>>>> >>> try to come up with a name for unnamed jars. It might look like the
>>>> >>> feature of automodules makes migrating easier because every dependency
>>>> >>> will get a name so can complete your module-info for all requirements,
>>>> >>> but
>>>> >>> we expect that once Jigsaw comes to speed the invalid module names are
>>>> >>> actually blocking further development due to name collisions or forced
>>>> >>> renaming by transitive modular jars.

>>>> >>> The advantage of this proposal is that library builders are not forced
>>>> >>> to
>>>> >>> keep the proposed module name in order to maintain backwards
>>>> >>> compatibility
>>>> >>> with the default.. Instead library builders can pick a more suitable
>>>> >>> module name. The modular system doesn’t allow the same package to be
>>>> >>> exported by multiple jars (and automodules exports every package).
>>>> >>> Library
>>>> >>> builders can fix this is their new jars, however if end users would
>>>> >>> require both jars because they were specified as requirements in
>>>> >>> different
>>>> >>> transitive jars, you cannot compile this project. There’s just no
>>>> >>> dependency-excludes like Maven has, because “requires” in the
>>>> >>> module-info
>>>> >>> really means requires. Dropping automodules will prevent these kind of
>>>> >>> issues, because a package can only be exported by a named module.

>>>> >>> Sure, this means that for end users they cannot refer to every jar in
>>>> >>> their module-info. But at least if they add a “requires” to their
>>>> >>> module-info, they can ensure that it’ll always refer to the intended
>>>> >>> modular jar. With build tools like Maven the chance of missing artifacts
>>>> >>> on the classpath has already been reduced a lot. In general builds have
>>>> >>> become quite stable, so we don’t expect that developers will translate
>>>> >>> all
>>>> >>> dependencies to the module-info file, especially if we warn them about
>>>> >>> the
>>>> >>> possible consequences of depending on automodules. Only referring to
>>>> >>> named
>>>> >>> modules and even a single “requires” is already a gain. There’s no
>>>> >>> reason
>>>> >>> to try to speed this up and give the developer the false impression that
>>>> >>> it’ll keep working when upgrading to real modular jars. Focus should be
>>>> >>> on
>>>> >>> the target, not on the path how to reach it.

>>>> >>> Dropping the automodules will prevent a lot of discussions about what is
>>>> >>> the correct way to select a module name and will give the responsibility
>>>> >>> for the name back to the place where it belongs: the developer.

>>>> >>> [1]
>>>> >>> http://stackoverflow.com/questions/22053381/lack-of-available-module-names-on-npm
>>>> >>> [2]
>>>> >>> http://blog.npmjs.org/post/116936804365/solving-npms-hard-problem-naming-packages
>>>> >>> [3] The fact that so much of the npm ecosystem is effectively
>>>> >>> not-namespaced is has actually
>>>> >>> created potential build time malware injection possibilities. If I know
>>>> >>> of
>>>> >>> a package in use by a
>>>> >>> company through log analysis, bug report analysis etc, I could
>>>> >>> potentially
>>>> >>> go register the same
>>>> >>> name in the default repo with a very high semver and know that it’s very
>>>> >>> likely this would be
>>>> >>> picked up over the intended internally developed module because there’s
>>>> >>> no
>>>> >>> namespace.
>>>> >>> [4] https://en.wikipedia.org/wiki/Default_effect_(psychology)
>>>> >>> [5] https://en.wikipedia.org/wiki/Principle_of_least_astonishment
>>>> >>> [6]
>>>> >>> https://docs.google.com/spreadsheets/d/1TVR5uTpDYw0827AlvPRu8l95zHnFPL_g61TdPtnj
>>>> >>> Q5M/edit?usp=sharing
>>>> >>> [7] http://openjdk.java.net/jeps/261 #Risk and assumptions
>>>> >>> [8]
>>>> > >> https://www.mail-archive.com/jigsaw-dev@openjdk.java.net/msg06623.html