You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@solr.apache.org by Christos Malliaridis <c....@gmail.com> on 2024/08/12 14:33:12 UTC

Introducing Gradle Version Catalogs

In the Lucene project there was a couple weeks ago a migration to version
catalogs (see PR #13484 <https://github.com/apache/lucene/pull/13484/>).

I'd like to follow the same approach in Solr, as it would allow us to sync
some of the dependencies and keep all versions in one file.

If this is something to look forward to, I would create a ticket and a
first draft PR.

Sincerely,
Christos

Re: Introducing Gradle Version Catalogs

Posted by Jan Høydahl <ja...@cominvent.com>.
I propose we try dependabot first for the new main - benefit is we don't need to host it ourselves.
If version catalogs are not backported to 9x, perhaps we can re-purpose SolrBot to file PRs for branch_9_N?

Jan

> 14. nov. 2024 kl. 05:47 skrev Christos Malliaridis <ma...@apache.org>:
> 
> The current bot would require some configuration changes to get it running again, as renovate relies on palantirs plugin for updating lock files, and I believe it uses "gradlew --write-locks" to update the file. This would not work anymore.
> 
> The bot already runs a custom task for including "gradlew updateLicenses", so it should be possible to also include "gradlew writeLocks". I tried to run it with updated configs on a project clone, but failed to get it up and running (it fails to fork the project).
> 
> I reviewed debendabot as well and it seems quite easy to get it up and running. We will only have to add add a basic config file and a workflow for running the two tasks "gradlew writeLocks" and "gradlew updateLicenses", as these cannot be configured in dependabot from what I know.
> 
> I think dependabot comes with better GitHub integration and it would be worth considering a switch, as Jan proposed.
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
> For additional commands, e-mail: dev-help@solr.apache.org
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <ma...@apache.org>.
The current bot would require some configuration changes to get it running again, as renovate relies on palantirs plugin for updating lock files, and I believe it uses "gradlew --write-locks" to update the file. This would not work anymore.

The bot already runs a custom task for including "gradlew updateLicenses", so it should be possible to also include "gradlew writeLocks". I tried to run it with updated configs on a project clone, but failed to get it up and running (it fails to fork the project).

I reviewed debendabot as well and it seems quite easy to get it up and running. We will only have to add add a basic config file and a workflow for running the two tasks "gradlew writeLocks" and "gradlew updateLicenses", as these cannot be configured in dependabot from what I know.

I think dependabot comes with better GitHub integration and it would be worth considering a switch, as Jan proposed.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Jan Høydahl <ja...@cominvent.com>.
We’ll need to modify what files renovatebot considers, right now it is only two files, after it will probably be several.

Another option is to switch to Dependabot, which did not support consistent-versions plugin, but do support version catalogs.

Jan

> On 13 Nov 2024, at 16:38, David Smiley <ds...@apache.org> wrote:
> 
> Thanks again for this modernization effort, Christos!  A lot of work,
> clearly.
> 
> Do you imagine the "Renovate" bot will need updates to work in this new
> world?  I suppose we shall deal with that when we get to the other side.
> 
>> On Wed, Nov 13, 2024 at 10:12 AM Christos Malliaridis <
>> malliaridis@apache.org> wrote:
>> 
>> This is the last call for SOLR-17406 /
>> https://github.com/apache/solr/pull/2706. All matters have been resolved
>> and the changes will be merged soon if there are no objections.
>> 
>> If you are having any open PRs that make any changes in gradle files, I'd
>> recommend to wait before merging. If you add dependencies in your PRs, you
>> will likely have to update to the new dependency management format that
>> uses references instead of strings. This would be a great opportunity to
>> review the updated documentation in practice.
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
>> For additional commands, e-mail: dev-help@solr.apache.org
>> 
>> 

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by David Smiley <ds...@apache.org>.
Thanks again for this modernization effort, Christos!  A lot of work,
clearly.

Do you imagine the "Renovate" bot will need updates to work in this new
world?  I suppose we shall deal with that when we get to the other side.

On Wed, Nov 13, 2024 at 10:12 AM Christos Malliaridis <
malliaridis@apache.org> wrote:

> This is the last call for SOLR-17406 /
> https://github.com/apache/solr/pull/2706. All matters have been resolved
> and the changes will be merged soon if there are no objections.
>
> If you are having any open PRs that make any changes in gradle files, I'd
> recommend to wait before merging. If you add dependencies in your PRs, you
> will likely have to update to the new dependency management format that
> uses references instead of strings. This would be a great opportunity to
> review the updated documentation in practice.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
> For additional commands, e-mail: dev-help@solr.apache.org
>
>

Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <ma...@apache.org>.
This is the last call for SOLR-17406 / https://github.com/apache/solr/pull/2706. All matters have been resolved and the changes will be merged soon if there are no objections.

If you are having any open PRs that make any changes in gradle files, I'd recommend to wait before merging. If you add dependencies in your PRs, you will likely have to update to the new dependency management format that uses references instead of strings. This would be a great opportunity to review the updated documentation in practice.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <ma...@apache.org>.
Bringing up this topic again as the PR https://github.com/apache/solr/pull/2706 is ready for another review round.

Since it changes the way dependencies are managed, it would be great if we can get more eyes on the PR before merging it.

This is also the last "blocker" of the POC for the new Admin UI.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
Hi Christos,

I've now revised the initial PR (https://github.com/apache/solr/pull/2646)
> and created a clean PR with the changes and version conflict resolution.
> See for that https://github.com/apache/solr/pull/2706. The new PR does
> not include dependency updates like the initial PR.
>

Nice! I think it's a good start. Updates can come later, even one at a time.


> Right now, the constraints hold information about the dependencies that
> cause the version conflict, so that in the future developers may check the
> constraints of the updated dependency for matching constraints and clean it
> up if the constraint is no longer necessary. I tried to find an easier or
> more automated way to properly maintain the constraints that have been
> added, but I couldn't find a better one. So if you have any ideas how this
> could be improved, please let me know.
>

I think I mentioned this already but, sadly, I haven't found any more
"automated" way of managing these constraints along with version updates.
I'm sure one could try to craft something... but it's an
uncharted territory for me and it's never been a problem for me to just
review them manually.

Dawid

Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <c....@gmail.com>.
I've now revised the initial PR (https://github.com/apache/solr/pull/2646)
and created a clean PR with the changes and version conflict resolution.
See for that https://github.com/apache/solr/pull/2706. The new PR does not
include dependency updates like the initial PR.

Right now, the constraints hold information about the dependencies that
cause the version conflict, so that in the future developers may check the
constraints of the updated dependency for matching constraints and clean it
up if the constraint is no longer necessary. I tried to find an easier or
more automated way to properly maintain the constraints that have been
added, but I couldn't find a better one. So if you have any ideas how this
could be improved, please let me know.

>

Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
> There isn't a singular "classpath"
> (DependencySet? in Gradle).  I think there are multiple "classpaths"
> for many things (e.g. compiling main sourceSet vs compiling test
> sourceSet) for each module (many modules).


These configurations are from the java plugin in gradle and most of them
are tied together because of inheritance between configurations. But in
general - no - configurations can have any dependencies you wish (and they
are resolved within a configuration and its network of inherited
configurations and constraints). There is flexibility and beauty in it -
much like with different class loaders having different set of jars. But it
does get complicated when you're trying to ensure consistency across
different independent configurations.

Different artifact resolution from the same initial conditions (dependency
version and constraints) is easier to show for two different subprojects -
it is typically enough to have a transitive dependency at a different minor
version in one of them to resolve differently. Gradle has its own
recommendations on how you can solve this but they're beyond my
understanding (especially when they don't work as advertised...).


> But every classpath
> produced by Gradle (DependencySet) should have one version of a given
> dependency (group+artifactId).
>

Eh... It's a longer story... It's always a configuration resolution - there
is no global set of dependencies. Even palantir's plugin doesn't guarantee
consistent resolution, actually. That's why that plugin of mine is useful -
it picks the final artifacts and verifies they use unique version number
across the set of configurations you desire.

Dawid

Re: Introducing Gradle Version Catalogs

Posted by David Smiley <ds...@apache.org>.
On Thu, Aug 22, 2024 at 3:25 AM Dawid Weiss <da...@gmail.com> wrote:
> > What would be the drawbacks / risks for allowing multiple versions of the
> > same library in case of transitive dependencies and their resolution?
> >
>
> The drawback is that you'd end up having multiple JARs on the classpath -
> their ordering will affect class resolution, so probably not acceptable.

I wasn't expecting that.  There isn't a singular "classpath"
(DependencySet? in Gradle).  I think there are multiple "classpaths"
for many things (e.g. compiling main sourceSet vs compiling test
sourceSet) for each module (many modules).  But every classpath
produced by Gradle (DependencySet) should have one version of a given
dependency (group+artifactId).

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
>
> I couldn't find a way to identify redundant constraints either for
> cleaning up in the long run the constraints. Any ideas how this could be
> tackled?
>

Exactly. It's a problem I keep struggling with elsewhere as well. The only
workflow I managed to figure out was to keep the constraints centralized,
remove them upon a batch of upgrades, then selectively re-add those that
are still needed. Doing this manually would involve a lot of tinkering with
gradle's internals, I think... You could try to clone/recreate selected
configurations, then resolve them without constraints and compare the
results. I don't know how complex it'd get.


> What would be the drawbacks / risks for allowing multiple versions of the
> same library in case of transitive dependencies and their resolution?
>

The drawback is that you'd end up having multiple JARs on the classpath -
their ordering will affect class resolution, so probably not acceptable.

Is that even possible if we lock dependency versions?


Gradle will align versions of the same dependency (or complain) within a
single configuration but for two "independent" configurations, the same
catalog version may resolve differently depending on transitives. It's
mentioned here:

https://docs.gradle.org/current/userguide/resolution_strategy_tuning.html#resolution_consistency

and it's the reason gradle's java plugin provides odd hacks like this one:

https://docs.gradle.org/current/userguide/resolution_strategy_tuning.html#sec:java_consistency
https://docs.gradle.org/current/javadoc/org/gradle/api/plugins/JavaResolutionConsistency.html

It won't help if you have custom configurations that you want to keep
aligned - that's the reason I wrote that custom plugin...

Dawid

Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <c....@gmail.com>.
> What palantir-consistent-versions did (always taking the ceiling) is
easier but less transparent,
I am for transparency too, I had enough troubleshooting sessions for fixing
something that I couldn't see.

What I also found is that there are some difficulties in keeping the
constraints clean the way we plan to resolve the reported conflicts. Once a
constraint is added, another constraint may resolve as well, but you only
know of its redundancy if you run the gradle task inbetween each constraint
addition / removal, or check the dependency insights for each conflict
individually. I couldn't find a way to identify redundant constraints
either for cleaning up in the long run the constraints. Any ideas how this
could be tackled?

What would be the drawbacks / risks for allowing multiple versions of the
same library in case of transitive dependencies and their resolution? Is
that even possible if we lock dependency versions?

Christos

On Mon, Aug 19, 2024 at 8:17 PM Dawid Weiss <da...@gmail.com> wrote:

> Hi Christos,
>
> > I am just not very happy with the repetitive constraints and the need to
> include transitive dependencies in the version catalog whenever there is a
> version mismatch that needs to be synced.
>
> I share your pain - there is added verbosity here. I haven't found an
> elegant way to deal with this. What palantir-consistent-versions did
> (always taking the ceiling) is easier but less transparent, in my
> opinion. I look forward to seeing something that would be a nice
> middle ground.
>
> > I tried integrating a platform module to understand its use case and
> limitations, but it doesn't seem to be very helpful here. I therefore share
> your opinion on that. It would be something we could consider if we publish
> each Solr module separately and allow the users to sync the versions of
> each module with the platform (as a BOM probably).
>
> I did some prototyping with BOMs and platform modules but it never
> worked for me quite right (I played with some intentionally hairy
> transitive conflicts). Maybe it's my lack of experience or my lack of
> understanding that is to blame here.
>
> > In the other thread about Resolving dependencies with depdencychecks
> plugin and in the Lucene project you went for two configuration groups, one
> for main and one for test. I would probably do the same, but when I thought
> it through, I was not sure if this is what we want. I assume unexpected
> behavior if the main / compile and runtime configurations may use different
> versions of libraries than the test compile and test runtime.
>
> You are correct - the two different configurations (main/test) may
> resolve transitives in a different way. My reasoning was that it's an
> unlikely event - most of the time the test configuration will inherit
> a bulk of deps from the main one, then add test-only libraries. The
> overlap in their transitive set should be small or even none. The
> benefit of keeping the two separate is to have an "inventory list" in
> the lock file - you can clearly see when a new library (or an upgrade)
> made it to the main set of dependencies when you're updating the lock
> file and running git diff... This said, that plugin of mine is
> actually fairly agnostic in terms of the number or arrangement of
> configurations which are checked for consistency. So you could have
> three separate ones - main, test and main+test... If there is an
> inconsistency, it will be signalled and you'll have to correct it via
> the constraint mechanism. Again - not sure which way is the best way
> (what you did is likely better for consistency).
>
> Dawid
>

Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
Hi Christos,

> I am just not very happy with the repetitive constraints and the need to include transitive dependencies in the version catalog whenever there is a version mismatch that needs to be synced.

I share your pain - there is added verbosity here. I haven't found an
elegant way to deal with this. What palantir-consistent-versions did
(always taking the ceiling) is easier but less transparent, in my
opinion. I look forward to seeing something that would be a nice
middle ground.

> I tried integrating a platform module to understand its use case and limitations, but it doesn't seem to be very helpful here. I therefore share your opinion on that. It would be something we could consider if we publish each Solr module separately and allow the users to sync the versions of each module with the platform (as a BOM probably).

I did some prototyping with BOMs and platform modules but it never
worked for me quite right (I played with some intentionally hairy
transitive conflicts). Maybe it's my lack of experience or my lack of
understanding that is to blame here.

> In the other thread about Resolving dependencies with depdencychecks plugin and in the Lucene project you went for two configuration groups, one for main and one for test. I would probably do the same, but when I thought it through, I was not sure if this is what we want. I assume unexpected behavior if the main / compile and runtime configurations may use different versions of libraries than the test compile and test runtime.

You are correct - the two different configurations (main/test) may
resolve transitives in a different way. My reasoning was that it's an
unlikely event - most of the time the test configuration will inherit
a bulk of deps from the main one, then add test-only libraries. The
overlap in their transitive set should be small or even none. The
benefit of keeping the two separate is to have an "inventory list" in
the lock file - you can clearly see when a new library (or an upgrade)
made it to the main set of dependencies when you're updating the lock
file and running git diff... This said, that plugin of mine is
actually fairly agnostic in terms of the number or arrangement of
configurations which are checked for consistency. So you could have
three separate ones - main, test and main+test... If there is an
inconsistency, it will be signalled and you'll have to correct it via
the constraint mechanism. Again - not sure which way is the best way
(what you did is likely better for consistency).

Dawid

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <c....@gmail.com>.
I worked on the PR and solved the version conflicts now. I considered
removing the plugin, as you proposed, but I ended up using it instead,
because its functionality is still useful in my opinion. I am just not very
happy with the repetitive constraints and the need to include transitive
dependencies in the version catalog whenever there is a version mismatch
that needs to be synced. But I still prefer it this way instead of the way
Gradle resolves conflicts.

Some of the repetitiveness in the dependencies.gradle for resolving
conflicts could be simplified by just iterating over the libraries of the
catalog and adding them all as constraints. But I felt like doing so could
have unexpected resolutions, so I went for the repetitiveness.

> I've toyed with gradle's notion of platforms but they never worked the
way I wanted them to.

I tried integrating a platform module to understand its use case and
limitations, but it doesn't seem to be very helpful here. I therefore share
your opinion on that. It would be something we could consider if we publish
each Solr module separately and allow the users to sync the versions of
each module with the platform (as a BOM probably). But that seems to be
only relevant if the users integrate individual Solr modules into other
projects, and it doesn't look to me like Solr is that kind of project
(correct me if I'm wrong).

I also thought that applying a BOM from another project as a constraint
would resolve at least some of the conflicts, but I think that is not the
case.

I am now planning to improve the documentation for the dependencies,
especially the licensing part. It is quite time consuming and difficult if
you don't know where to search for the license and notice files.

In the other thread about Resolving dependencies with depdencychecks plugin
<https://lists.apache.org/thread/mhvcfxy6dh2x1zqbxvrnjqch47222rjl> and in
the Lucene project you went for two configuration groups, one for main and
one for test. I would probably do the same, but when I thought it through,
I was not sure if this is what we want. I assume unexpected behavior if the
main / compile and runtime configurations may use different versions of
libraries than the test compile and test runtime. Since the tests are
supposed to ensure the stability of the project with the libraries
selected, I went for a group where both main and test configurations (from
your example) are included, so that when executing tests, the same library
versions are used as in the final build artifact. I don't know if this way
of thinking is actually correct, so any thoughts on that would be very
helpful.

Best,
Christos

>

Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
> Indeed, I saw that you also added spotless and tidy to the gradle build
> files and provided a gradle plugin for the dependency checks. I definitely
> and unintentionally talked down your work in that PR, sorry for that.
>

That's fine. I'm not that happy with everything that's in the Lucene build
right now. It started much simpler, then - as usual - grew tentacles all
over. Solr is going to be even more challenging because of the sheer number
of dependencies. But maybe I'm wrong - no other way to test this other than
by trying.


> So I applied most of the changes on top of mine and have almost finished
> the migration. I just fail on some dependency resolution constraints that
> probably need to be provided manually. I would appreciate any input on that
> (I'll create a draft PR soon).
>

I've sent a separate e-mail about this - sorry, didn't see this thread
until now. In short: I don't have any bright ideas other than what I have
used in the past. I've toyed with gradle's notion of platforms but they
never worked the way I wanted them to. It was probably me not understanding
their concept. What I did for Lucene (custom version validation plugin +
resolution constraints + not much else) is what I have been using
elsewhere. I can understand what it does but I agree with you that it'd be
good to improve on the documentation. Some guidelines for Lucene devs are
here:

https://github.com/apache/lucene/blob/main/help/dependencies.txt#L34-L82

Dawid

Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <c....@gmail.com>.
I've created SOLR-17406 <https://issues.apache.org/jira/browse/SOLR-17406>
 and PR 2646 <https://github.com/apache/solr/pull/2646> so that we have
something to work with. We probably have more than enough time to finish
this migration before Solr 10 is released.

Re: Introducing Gradle Version Catalogs

Posted by David Smiley <ds...@apache.org>.
Christos, since you've done seemingly all the work or close to it, IMO
it's better to continue instead of delaying merging to main.

It's tempting to respond to Solr 10 & 9 releases and timelines but
that is deserving of another thread that would derail this thread IMO.
Or chit-chat in Slack on that; it's not decision making.

On Wed, Aug 14, 2024 at 6:45 AM David Eric Pugh <de...@yahoo.com.invalid> wrote:
>
>  My understanding is that Lucene 10 is likely to be released this fall, and I know I'd like to see Solr 10 come out close to when Lucene 10 comes out so we don't fall behind....   So maybe there won't be many more releases of 9?  9.8 maybe?  but not a 9x line that gets into the double digits on the minor version?
>
>     On Tuesday, August 13, 2024 at 04:54:59 PM EDT, Christos Malliaridis <c....@gmail.com> wrote:
>
>  > It's not just that.
>
> Indeed, I saw that you also added spotless and tidy to the gradle build
> files and provided a gradle plugin for the dependency checks. I definitely
> and unintentionally talked down your work in that PR, sorry for that.
>
> I started a couple weeks ago a migration without knowing that there was a
> similar move in Lucene, until David mentioned the Lucene PR. So I applied
> most of the changes on top of mine and have almost finished the migration.
> I just fail on some dependency resolution constraints that probably need to
> be provided manually. I would appreciate any input on that (I'll create a
> draft PR soon).
>
> > We might want to delay such a change for another release or so in order for
> dependency backports to 9 to go smoothly.  There's no hurry.
>
> Definitely. I just wanted to confirm the plans and share my current
> progress in case it is the way to go, which looks like it. I'll file an
> issue then so that we can keep track of that. Introducing the migration in
> Solr 10 sounds also like the right moment to me.
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org


Re: Introducing Gradle Version Catalogs

Posted by David Eric Pugh <de...@yahoo.com.INVALID>.
 My understanding is that Lucene 10 is likely to be released this fall, and I know I'd like to see Solr 10 come out close to when Lucene 10 comes out so we don't fall behind....   So maybe there won't be many more releases of 9?  9.8 maybe?  but not a 9x line that gets into the double digits on the minor version?

    On Tuesday, August 13, 2024 at 04:54:59 PM EDT, Christos Malliaridis <c....@gmail.com> wrote:  
 
 > It's not just that.

Indeed, I saw that you also added spotless and tidy to the gradle build
files and provided a gradle plugin for the dependency checks. I definitely
and unintentionally talked down your work in that PR, sorry for that.

I started a couple weeks ago a migration without knowing that there was a
similar move in Lucene, until David mentioned the Lucene PR. So I applied
most of the changes on top of mine and have almost finished the migration.
I just fail on some dependency resolution constraints that probably need to
be provided manually. I would appreciate any input on that (I'll create a
draft PR soon).

> We might want to delay such a change for another release or so in order for
dependency backports to 9 to go smoothly.  There's no hurry.

Definitely. I just wanted to confirm the plans and share my current
progress in case it is the way to go, which looks like it. I'll file an
issue then so that we can keep track of that. Introducing the migration in
Solr 10 sounds also like the right moment to me.
  

Re: Introducing Gradle Version Catalogs

Posted by Christos Malliaridis <c....@gmail.com>.
> It's not just that.

Indeed, I saw that you also added spotless and tidy to the gradle build
files and provided a gradle plugin for the dependency checks. I definitely
and unintentionally talked down your work in that PR, sorry for that.

I started a couple weeks ago a migration without knowing that there was a
similar move in Lucene, until David mentioned the Lucene PR. So I applied
most of the changes on top of mine and have almost finished the migration.
I just fail on some dependency resolution constraints that probably need to
be provided manually. I would appreciate any input on that (I'll create a
draft PR soon).

> We might want to delay such a change for another release or so in order for
dependency backports to 9 to go smoothly.  There's no hurry.

Definitely. I just wanted to confirm the plans and share my current
progress in case it is the way to go, which looks like it. I'll file an
issue then so that we can keep track of that. Introducing the migration in
Solr 10 sounds also like the right moment to me.

Re: Introducing Gradle Version Catalogs

Posted by Dawid Weiss <da...@gmail.com>.
Hi Christos,

I'd like to follow the same approach in Solr, as it would allow us to sync
> some of the dependencies and keep all versions in one file.
>

It's not just that. It's also a switch from palantir's dependency
resolution mechanism to gradle's.
I think it's a good direction, because palantir's consistent-versions
plugin is all-or-nothing,
whereas with gradle's built-in mechanism is much more flexible, but also
less intuitive. The same dependency
catalog version can resolve to a different version depending on the
configuration context...
this can be mind-boggling at times (at least to me).

I share David's opinion that this is going to be a bigger change than it
appears on the surface. If you need anything,
please CC me directly as I am no longer that involved in Solr development.

Dawid

Re: Introducing Gradle Version Catalogs

Posted by David Smiley <ds...@apache.org>.
Thanks for offering to help port this!

I noticed that change is Lucene 10 only; I think a build change of
that magnitude should also be Solr 10 only, thus the main branch.  We
might want to delay such a change for another release or so in order
for dependency backports to 9 to go smoothly.  There's no hurry.

On Mon, Aug 12, 2024 at 10:34 AM Christos Malliaridis
<c....@gmail.com> wrote:
>
> In the Lucene project there was a couple weeks ago a migration to version
> catalogs (see PR #13484 <https://github.com/apache/lucene/pull/13484/>).
>
> I'd like to follow the same approach in Solr, as it would allow us to sync
> some of the dependencies and keep all versions in one file.
>
> If this is something to look forward to, I would create a ticket and a
> first draft PR.
>
> Sincerely,
> Christos

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org