Page Comparison

Jenn Colt  is next, followed by Florian Gleixner, then Jeremy Huff 

Reminder:  Please copy/paste the Zoom chat into the notes.  If you miss it, this is saved along with the meeting recording, but having it here has benefits. 

Florian Gleixner 

Present abstract for RFC:  https://github.com/folio-org/rfcs/pull/23

Florian Gleixner : This RFC is about configuration options in FOLIO. It describes where to put configurations. The goal is to drop mod configuration due to security. It proposes favoring distributed configuration over the centralized configuration. It proposes that mod-configuration should not be used until the R release. The RFC suggests that a some exceptions might be made for centralized configuration. The RFC lays out a timing for migration. 

Tod Olson: Central configuration could be used in cases where modules do not have storage. E.G. z39.50 or other edge modules.

Craig McNally : That is a good point, and the UI consumes these configurations.

Florian Gleixner : There is no module that could be the parent of these settings, so these can be placed in a centralized configuration module. 

Tod Olson : will there be guidelines to promote consistency in how settings are expressed.

Florian Gleixner : this was discussed with Julian Ladisch and they decided that this was out of scope.

Tod Olson : So these changes are considered to be near-term

Florian Gleixner : This is true because of the security concerns. Maybe in a future RFC guidance could be given for best practices for designing settings.

Craig McNally : put forward a question from the chat, from Marc Johnson: How is the scope near-term, he thought this was intended to be a long term strategic decision.

Florian Gleixner : Dropping mod configuration is near term, but the use of distributed configuration is a more long term strategic decision. 

There were no objections and the RFC moved forward

Jakub Skoczen / Mike Taylor 

Present abstract for RFC:  https://github.com/folio-org/rfcs/pull/25 

Jakub Skoczen : This RFC proposes that GO be added to the officially supported technology list for backend development. Go is a good language for building web micro services. It is a simple language and they feel it is a good fit. The timing sections suggests that a pilot module be created to determine the long term suitability of the language. If it is not successful GO will be dropped from the officially supported technologies list.

Ingolf Kuss : Is there already a plan for a pilot module

Jakub Skoczen : Yes there is a plan for a a pilot module. He imagines the module would be a fairly fringe module.

Marc Johnson : He has thoughts that this will have a big president  setting effect, in terms of the polyglot nature of FOLIO

Jakub Skoczen : They had the thought that this discussion might have a more general component, rather than just being about the specifics of Go.

Jeremy Huff : How much will this RFC dive into containerization?

Jakub Skoczen : That could be a direction that the conversation could take. He feels that containerization would be a less controversial aspect of the conversation, and that maintenance of a new code base would be a more robust topic.

Craig McNally : He cautions us against adding the language to the officially approved technology list provisionally.

There were no objections and the RFC moved forward

Some items on this page are accompanied by specific versions or version ranges, while others are not.  Discuss whether or not we want to make this more consistent, or at least make it more clear when a version should be specified.  The primary concern likely has to do with how these pages are expected to be used.  There are at least a few different use cases includes, but not limited to:  

• During the TCR process to verify that the module under review isn't introducing new technologies, or incompatible versions of technologies already on the list.
• To track/plan upgrades of dependencies to ensure we're using supported versions (security implications)
• ...

Marc Johnson : There could be timing issues with our discussion of the Officially Approved Technology list and the two previously proposed RFCs

Jeremy Huff : Do we need this list to be so granular, for instance would both the spring module core and spring way need to be enumerated on the list

Craig McNally : This is most likely out of scope, but it seems like if they both use the same version of spring, would this ok

Marc Johnson : There are historical issues with this that are worth talking about

Craig McNally : That is true, and this should be added to the topic backlog. Todays discussion is about versioning. Should these pages maintain versioning?

Jeremy Huff : This list was created for module evaluations. Its use in support periods was added later. He is of the opinion that it should not be used for both purposes. 

Marc Johnson : If we divided these into multiple lists, they would necessarily be related. How would we handle dividing it into multiple lists

Jeremy Huff : We should remove versions from the Officially Approved Technology list and maintain supported versions in another document

Marc Johnson : There are issues with these lists going out of sync and presenting logical contradictions. For the things that are shared (i.e. stripes), the versions mater. 

Jakub Skoczen : The team that is using  software that is out of support is on the hook for making the changes. We want the version to be as high as is possible.

Marc Johnson : He disagrees that we always want to be on the highest version of everything. Even if the goal is to get to the latest we still have to describe the steps as to how we get there.

Tod Olson : We started this list, and then it became apparent that in some cases versions matter. He agrees that some of the things that are shared (like

postgres) are different, but the versions do need to be tied to a FOLIO version. He feels that he is broadly supporting what Marc Johnson is suggesting, and feels that it would be awkward separating these into multiple documents.

Jakub Skoczen : He agrees with Tod Olson , we should probably express somewhere that we want versions to progress as quickly as is possible. We should not optimize for situations that are holding us back, but instead embrace the goal that we are using the most recent dependencies.

Craig McNally : We are all agreeing that we want to be moving forward, and avoid using deprecated software. We are running out of time, but it does sound like it is

worth capturing the idea that we only specify versions "where they matter", as in instances that are cross-cutting.

Marc Johnson : These are dependencies that are shared between multiple dev teams, or where the system won't work if these elements are not aligned.