freemarker-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Siegfried Goeschl <>
Subject Re: FreeMarker generator release preparations
Date Mon, 10 Aug 2020 18:53:46 GMT
Hi Daniel,

Please see my comments below

Thanks in advance, 

Siegfried Goeschl

> On 09.08.2020, at 22:00, Daniel Dekany <> wrote:
> See my answers inline...
> On Sun, Aug 9, 2020 at 7:16 PM Siegfried Goeschl <
> <>> wrote:
>> Hi Daniel,
>> I think there is a way to define a custom FreeMarker template path in
>> IntelliJ - see
>> Having said that we should take care that the IntelliJ plugin is happy
>> * Some users do create web applications based on Apache FreeMarker
>> * Having FTL support built-in is a huge bonus for beginners
> What do you mean, realistically what can we do, in case that plugin is
> indeed that broken? I mean, for FreeMarker Generator alone it doesn't
> really matter for users. For all the other projects, FreeMarker doesn't
> prevent them from doing dirty things to accomodate to the plugin (as you
> saw yourself ;) ), and I'm not sure what else could FreeMarker itself do
> about this. Fixing/extending the plugin itself is also not likely, as it's
> closed source. (Of course, it would be great to have an open source LSP
> language server, and then adding open source plugins based on that... but
> things go very, very slowly around here.)

Played around with the IntelliJ plugin but I don't get it perfectly working ...
> Regarding the intended role of "/templates"
>> * The implemented template loading tries to resolve the given template
>> name as file and if there is no file the actual template loading is
>> delegated to a "MultiTemplateLoader"
>> * In I split the
>> existing templates into a reusable part (living now in "templates") and
>> examples (living now in "examples")
>> * IMHO the "templates" directory should contain commonly useful templates
>> - so they could be considered as guaranteed functionality
>>        * Transforming CSVs
>>        * Transforming Excel files to CSV, HTML & Markdown (and maybe
>> Confluence Markup)
>>        * Transform JSON to YAML and the other way around
>> * I'm happy to have those templates as files in the installation
>> directories - better visibility and user can add there own templates (if
>> the want to play nasty)
> The fundamental problem with that is this. Currently, if you pull in
> freemarker-generator-cli as Maven dependency, the templates will not be
> available. Surely, because it's the CLI, you could say that it's not
> supposed to be used without the FreeMarker Generator Home Directory created
> somewhere, which contains the launch script and templates/ and all. But, if
> these templates are guaranteed functionality in FreeMarker Generator, then
> they don't strictly belong to the CLI. When we will have a proper Maven
> plugin for example, they should be still accessible. In that  case you only
> have your Maven dependencies, so the templates must come from there.
> Regarding visibility, it's a bit like with Java. Java classes are not too
> readable without looking at the source code either. That's not an advantage
> when it comes to "visibility", sure. But luckily this is open source, and
> it's very easy to get to the source code, if someone really cares (like
> from the Maven source artifact). That applies to core stuff implemented in
> FTL as well. So, the previously mentioned advantage (that they are
> available from a plain dependency) certainly overweights this disadvantage
> (less visibility).
> I don't think we want users to add their own templates to /templates, as
> they will be potentially lost when someone updates FreeMarker Generator,
> and is confusing for colleagues, etc. If we can prevent bad practices,
> let's do it. Especially as we also provide a cleaner
> solution, ~/.freemarker-generator/ added to the template root (which will
> change to ~/.freemarker-generator/templates now, as you say below). If
> people need a system-wide one, we can also add /etc/freemarker-generator,
> or something like that.

I currently disagree here - I like the visibility aspect and it is pretty difficult to get
rid of templates loaded from the classpath.

> What I want to change
>> * Use "templates" as FreeMarker template root for the CLI installation and
>> "~/.freemarker-generator" directory
>> * Adopt the examples and documentation, e.g.  "freemarker-generator -t
>> info.ftl" instead of "freemarker-generator -t templates/info.ftl"
> Except, we should avoid adding info.ftl and such to the template directory
> root. Not only because of potential name clashes, but because it can be
> confusing if someone looks at that (where's info.ftl coming from). How
> about reserving the "freemarker-generator" directory on the top level, and
> then it's  "freemarker-generator -t freemarker-generator/info.ftl".
> (Hopefully that helps realizing that it's something built into FreeMarker
> Generator.) Actually, maybe it should be even
> "freemarker-generator:/info.ftl ".

Mhmm, already pushed the changes :-(

Need to think about it

> Thanks!
> Daniel Dekany
>> What do you think?
>> Siegfried Goeschl
>>> On 05.08.2020, at 13:28, Daniel Dekany <> wrote:
>>> Ah, you have the FreeMarker plugin in IntelliJ, and that assumes
>> something
>>> crazy like that? The Community Edition doesn't have that plugin, and I
>>> never saw it. But if that plugin indeed assumes that the template root is
>>> the project root... that wouldn't make sense, in any real world project.
>>> Even if some have a "templates" directory laying around in the directory
>> as
>>> the POM, then that would be the template root (means, a template path is
>>> like "/foo.ftl", not "/templates/foo.ftl"), not the whole project. So, I
>>> certainly don't think we should align with a broken plugin, especially as
>>> most users won't be affected (they just use the binary).
>>> Also, I'm a bit confused about the intended role of "/templates". Is it
>>> core (guaranteed) functionality in Generator CLI? As opposed to a core
>>> functionality of Generator itself? (Sure, for now Generator can be called
>>> via CLI only, but you see what I mean.) Or is it just some examples
>> laying
>>> around, whose presence users shouldn't count on in their projects? If
>> it's
>>> core functionality, then it certainly should be a Java resource, packed
>>> into the jar.
>>> Also let's not skip the question from the earlier mail, regarding what
>> path
>>> should this be available for the TemplateLoader (and for -t)?
>>> On Wed, Aug 5, 2020 at 11:47 AM Siegfried Goeschl <
>>>> wrote:
>>>> Hi Daniel,
>>>> When moving the "templates" to a subdirectory IntelliJ complains about
>> not
>>>> finding the FTL include files - technically it is working but without
>>>> tweaking IntelliJ it spits out errors. And the Travis tests failed
>> while my
>>>> local tests succeeded - see
>>>> Regarding template root directory - I guess there is some room for
>>>> improvements
>>>> * I'm passing a list of template root directories (current directory,
>> CLI
>>>> installation directory, ~/.freemarker-cli/"
>>>> * The "templates" directory is just the way I organised the overall file
>>>> structure
>>>> * Additional template root directory can be passed on the command line
>>>> * The implementation can pick up any FTL file using absolute or relative
>>>> file paths (and URLs)
>>>> Thanks in advance,
>>>> Siegfried Goeschl
>>>>> On 04.08.2020, at 22:12, Daniel Dekany <>
>> wrote:
>>>>> Can you clarify, what part of IntelliJ is happy about that, and why?
>>>>> Also it reminds me of another thing I wrote about earlier. Generally,
>> we
>>>>> say that the template root directory shouldn't not contain anything but
>>>>> templates, so if the project root is the template root directory, or
>> the
>>>>> freemarker-generator home directory, that's a bit fishy. Yes, the
>> reason
>>>>> for this purist approach is partially security concerns with web
>>>>> applications, and this is not one. But whatever the original reasons
>> are,
>>>>> FreeMarker has this omnipresent concept of the template root directory,
>>>>> which is kind of a virtual root directory for a template (i.e., they
>> can
>>>>> leave it, and it's all templates, or rarely files referred from
>>>> templates,
>>>>> like images).
>>>>> Also, it's strange/confusing if the template root directory has a
>>>>> "templates" subdirectory. I mean, the template root directory is what
>> one
>>>>> would normally call the templates directory. As a template author, I
>>>> would
>>>>> expect some path like "/freemarker-generator/whatever.ftl" (also note
>>>> that
>>>>> it's an absolute path). There,  "/freemarker-generator/" tells me that
>>>> this
>>>>> is something included in freemarker-generator, and not templates of my
>>>>> project.
>>>>> On Tue, Aug 4, 2020 at 8:51 PM Siegfried Goeschl <
>>>>>> wrote:
>>>>>> I'm not happy about it :-)
>>>>>> * A top-level "templates" directory makes IntelliJ happy when includes
>>>> are
>>>>>> used
>>>>>> * I have an ExamplesTest which tests each documented command line
>> usage
>>>>>> and I prefer to have it in sync with the documentation
>>>>>> * I actually tried it but reverted the changes since the Travis build
>>>> fails
>>>>>> Thanks in advance,
>>>>>> Siegfried Goeschl
>>>>>>> On 28.07.2020, at 22:34, Daniel Dekany <>
>>>> wrote:
>>>>>>> It's just that "templates" belongs to the "main" source code
of the
>>>>>>> freemarker-generator-cli, similarly to
>>>> config/,
>>>>>>> which is under src/main. (I think many of us expect a Maven project
>>>> root
>>>>>>> directory to only contain the "noise" needed for build and legal,
>>>>>> "src"
>>>>>>> that contains *all* the essential source code.) Also, as the
>>>> tree
>>>>>>> doesn't mirror the binary distribution directory tree anyway,
so I'm
>>>> not
>>>>>>> sure if there's any fundamental reason to make an exception with
>>>>>>> "templates", and put it under the root.
>>>>>>> On Tue, Jul 28, 2020 at 4:09 PM Siegfried Goeschl <
>>>>>>>> wrote:
>>>>>>>> Hi folks,
>>>>>>>> Back from the mountains and in front of the keyboard again
>>>>>>>> * I created a JIRA ticket and will push a feature branch
soon (bad
>>>>>> habits
>>>>>>>> die hard)
>>>>>>>> * I will go through the licences (review and collect in "licenses"
>>>>>>>> directory)
>>>>>>>> * Good point about plugin versions being defined in apache
>> will
>>>>>>>> rework them
>>>>>>>> * I will delete the existing configuration of the "release"
>>>>>>>> Some things to discuss
>>>>>>>> * What are the benefits to move "templates" to "src/main/templates"?
>>>>>>>> Currently they are in sync with "freemarker-cli" which is
quite nice
>>>> for
>>>>>>>> tests & documentation ...
>>>>>>>> Thanks in advance,
>>>>>>>> Siegfried Goeschl
>>>>>>>>> On 26.07.2020, at 01:27, Daniel Dekany <>
>>>>>> wrote:
>>>>>>>>> I said I will help in the Apache release process, so
only focusing
>> on
>>>>>>>> that,
>>>>>>>>> so some points:
>>>>>>>>> - We are required to have a so-called source release
(every other
>>>>>>>>> artifact is optional in the policy). As we are using
>>>>>>>> org.apache:apache
>>>>>>>>> parent, that should generate that automatically, with
.asc and
>> sha512
>>>>>>>> and
>>>>>>>>> all. But currently it doesn't, because maven-release-plugin
>>>>>>>> config/argument
>>>>>>>>> is overwritten with this:
>>>>>>>> <arguments>-Dmaven.javadoc.skip=true</arguments>.
>>>>>>>>> We should keep configuring release at minimum, to avoid
>>>>>> accidents.
>>>>>>>>> Maybe as in
>>>>>>>>> - I assume we also want a binary release, for the CLI
only, and
>>>>>>>>> freemarker-generator-cli-x.y.z-*app*.zip (note the "-app")
will be
>>>> our
>>>>>>>>> binary release artifact. Then:
>>>>>>>>> - It bundles some dependency binaries that are not under
>>>> license.
>>>>>>>>>  Unfortunately, the licenses of those must be included
in the
>>>>>>>>> distribution.
>>>>>>>>>  See the LICENSE at
>>>>>> At
>>>>>>>>>  the bottom, it lists the licenses, then it refers to
the actual
>>>>>>>> license
>>>>>>>>>  files. As we will have many licenses, let's create a
>>>>>>>> directory
>>>>>>>>>  for them. (In the future, the dependencies have to be
>>>>>>>>> for changes.
>>>>>>>>>  Even version upgrades my pull in sneaky transient dependencies.
>>>>>> Some
>>>>>>>>>  licenses are not even allowed, so anything but ASL2,
>>>>>>>>>  BSD-without-advertisement-clause, will need closer attention.)
>>>>>>>>>  - I noticed that the documentation is not included in
the binary
>>>>>>>>>  distribution. But because of the extra legal burden
including it
>>>>>>>> would
>>>>>>>>>  bring (we have fonts and icons under CC-SA and SIL OFL
in the
>>>>>> Docgen
>>>>>>>>>  output), I actually prefer that to stay like that.
>>>>>>>>>  - .sha512 file is not yet generated
>>>>>>>>> - freemarker-generator-cli/src/site: If you agree, instead
of this
>> I
>>>>>>>>> will create freemarker-generator*-site*/src/docgen, and
convert the
>>>>>>>>> Markdown to XDocBook. For now this will be only the CLI
>>>> documentation,
>>>>>>>> and
>>>>>>>>> the JavaDoc, as the freemarker-generator-maven-plugin
is not ready.
>>>>>> One
>>>>>>>>> annoyance I realized is that we should have Docgen in
Maven Central
>>>>>>>> for the
>>>>>>>>> builds to work reliably in the future, which means that
Docgen has
>> to
>>>>>>>> be
>>>>>>>>> officially released (it never was, it's an internal tool).
>> would
>>>>>>>> be a
>>>>>>>>> minimalistic release, means, no announcement, no web
site, just the
>>>>>>>> bare
>>>>>>>>> minimum (i.e., source release, and deployment to Maven
Central). I
>>>>>> have
>>>>>>>>> some backlog there (Google keeps nagging me about mobile
>> but
>>>>>> I
>>>>>>>>> hope I can fix that in the coming days, then go through
>> official
>>>>>>>>> release process (takes 1-2 weeks).
>>>>>>>>> - Some smaller things:
>>>>>>>>>  -
>>>>>>>>>  - Having a "release" profile is also hopefully unnecessary,
>>>> because
>>>>>>>>>  org.apache:apache takes care of signing.
>>>>>>>>>  - We should also remove most plugin version management,
as many
>> of
>>>>>>>>>  those versions are set in org.apache:apache.
>>>>>>>>>  - freemarker-generator-cli/templates should be inside
>>>>>>>>>  freemarker-generator-cli/src/main/templates, I guess.
>>>>>>>>> P.s.: Siegfired asked our opinions in another thread.
I did my
>> part,
>>>>>> even
>>>>>>>>> too much (;, so, would be good if others participate
in that as
>> well.
>>>>>>>>> --
>>>>>>>>> Best regards,
>>>>>>>>> Daniel Dekany
>>>>>>> --
>>>>>>> Best regards,
>>>>>>> Daniel Dekany
>>>>> --
>>>>> Best regards,
>>>>> Daniel Dekany
>>> --
>>> Best regards,
>>> Daniel Dekany
> -- 
> Best regards,
> Daniel Dekany

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message