Hi all, It's been almost 3 weeks since the last message on this thread. I think it's worth picking up and acting on.
To summarize the discussion up to this point: * I see a few folks did a "+1" to the option A with a short grace period, which we might could argue has already come and passed. * Eric mentioned an LTS version that won't get new features but will get security updates. * Richard reframed it as a capacity problem and that OpenNLP has a small team it's difficult to stay on top of older releases. Did I capture those points accurately? Did I miss any perspectives? Thanks, Jeff On Wed, Jul 8, 2026 at 9:29 AM Martin Wiesner <[email protected]> wrote: > > Hi all, > > any other opinions / perspectives from the committers / PMC members? > I’d like to encourage everyone to participate in this important discussion. > > Please share those within the next 1 or 2 weeks so we can find a consensus > and come to a decision. > > Best > Martin > -- > > > Am 20.06.2026 um 12:38 schrieb Kristian Rickert <[email protected]>: > > > > Hi all, > > > > I agree with Richard's points. To add my two cents: > > > > It is increasingly important to keep software up to date. The higher > > velocity required to patch CVEs today (which is a positive development) > > results in significantly more churn when back-porting compared to the past. > > > > I empathize with the time-consuming nature of updating libraries; this > > empathy is reflected heavily in the 3.0 API. We implemented additional > > code specifically to ensure the transition from version 2 to 3 is as > > seamless as possible. We did this because we know maintaining backward > > APIs is hard. > > > > Furthermore, it is typical for projects to EOL two versions back when > > releasing a major version. This transition often happens even faster for > > projects with heavy dependencies. > > > > If anyone has any issues transitioning off a 1.x build, they can try the > > dev forums for help - I'm sure they'll find the guidance they need. > > > > Best regards, > > > > Kristian Rickert > > > > > > On Fri, Jun 19, 2026 at 1:52 PM Richard Zowalla <[email protected]> wrote: > > > >> In contrast to Solr, the OpenNLP community is very small, and backporting > >> and maintaining different branches (3.x, 2.x, 1.x) is time consuming and > >> honestly not much fun, unless you happen to have a day job that actually > >> needs it. > >> > >> So from my POV this is really a volunteer-capacity question. Also for > >> reviewing, testing and producing an actual release. > >> > >> If someone is willing to step up and do the work: fine. And consumers who > >> still need 1.9.x are free to fork it, or take a patch-based approach (like > >> it can be done via the TomEE patch plugin) to update the jars in place, > >> especially if this is solely about satisfying CVE scanners and similar. > >> We’ve been running an EOL policy in Apache TomEE for quite a while now, and > >> so do Apache Storm and Apache Tomcat. I don't think maintaining versions > >> endlessly is a good thing (at least from a volunteer capacity side of > >> things; you can still wrap around a business model . > >> > >> What happens when the next transitive CVE shows up in a lib that itself is > >> no longer patched? From my experience, that just gets cumbersome. > >> > >> We had that scenario in TomEE 9.x once Tomcat 10.0.x went EOL pretty fast; > >> so I was back porting Tomcat patches and patching inline for quite a few > >> months (wasn’t fun at all). > >> > >> In a perfect world with lots of people willing to do the work: fine. But > >> that doesn't apply to the current OpenNLP community IMHO, or at least not > >> yet. > >> > >> Richard > >> > >> > >> > >>> Am 19.06.2026 um 16:14 schrieb Eric Pugh < > >> [email protected]>: > >>> > >>> I am not a committer, but I’d leave towards leaving it open ended if the > >> tooling to produce a release remains working. > >>> > >>> I think EOL is very important to convey “hey, we no longer can produce a > >> release” or “we no longer have the in house knowledge to maintain this”, > >> but if the release process is still manageable, and people aren’t trying to > >> jam in NEW FEATURES into 1.x, then I don’t see why you need to close it > >> off. > >>> > >>> I am very appreciative of the 1.9.5 coming out, and I would hope that if > >> more CVE’s pop up, being able to publish a 1.9.6 would be great. > >>> > >>> > >>> What if, and this is just an idea, you reframed things? Instead of > >> talking about EOL, what if you talked about LTS: Long Term Support. > >>> > >>> 1.9 is our LTS. If a CVE pops up, you can expect a 1.9.6 or 1.9.7. > >> There will never be a 1.10 with new features. All new features will go > >> to OpenNLP 3. We reserve the right to decide when 1.9 line is no longer > >> LTS. > >>> > >>> > >>> My experience in Solr is that there is a HUGE set of people who are > >> happy with their specific solution, and won’t ever upgrade till there is a > >> big event. For them, knowing they are on a LTS version, and knowing that > >> it can be produced reasonably easily, seems like a win-win for everyone. > >> When 1.9.x becomes a pain to release, then call the LTS period “done”. > >>> > >>> I wrote down some specifics that I haven’t actually shared with the Solr > >> community yet, but here you go: > >> https://docs.google.com/document/d/17qJIfbSoRYvwrPt5OWjqmliwghflDzV9fA0XmXmNSBE/edit?usp=sharing > >>> > >>> Eric > >>> > >>> > >>>> On Jun 18, 2026, at 12:39 PM, Kristian Rickert <[email protected]> > >> wrote: > >>>> > >>>> a) +1 > >>>> b) I'd lean on a short grace period. > >>>> > >>>> > >>>> On Wed, Jun 17, 2026 at 11:52 PM Richard Zowalla <[email protected]> > >>>> wrote: > >>>> > >>>>> a) +1 > >>>>> b) b2/b3 (if other CVEs are approaching) > >>>>> > >>>>> Examples: https://tomcat.apache.org/tomcat-9.0.x-eos.html < > >> https://tomcat.apache.org/tomcat-9.0.x-eos.html> > >>>>> > >>>>>> Am 18.06.2026 um 05:23 schrieb Martin Wiesner <[email protected]>: > >>>>>> > >>>>>> Hi all, > >>>>>> > >>>>>> given recent security fixes that landed in OpenNLP's main branch and > >> the > >>>>> request for back porting these changes to the very oldskoolish 1.9.x > >> line > >>>>> [1], the people involved noticed that the efforts to maintain three > >>>>> separate version lines along the road were pretty high and resource > >>>>> consuming. > >>>>>> > >>>>>> Therefore, I'd like to propose to (finally) declare Apache OpenNLP > >> 1.9.x > >>>>> EOL publicly via a News announcement on the project's website. > >>>>>> > >>>>>> Primary questions: > >>>>>> (a) Do we have consensus that such an EOL announcement is long overdue > >>>>> and should be put out rather soonish? > >>>>>> (b) Time of the announcement: Options that I see: > >>>>>> - b1: Directly with the projected release of the 1.9.5, marking it as > >>>>> the last release ever to be expected for OpenNLP 1.x. > >>>>>> - b2: Shortly after - with a grace period - for instance End of July > >>>>> 2026, or similar short ranged targets. > >>>>>> - b3: End of year 2026, that is Dec 31, 2026 > >>>>>> (c) Are there any requirements by the ASF to put out an EOL > >>>>> announcement? Jeff, do you have infos about it? > >>>>>> > >>>>>> Open for others to add thoughts and related aspects to this > >> discussion. > >>>>>> Please share your opinions and provide (your) answers to question (a) > >> to > >>>>> (c). > >>>>>> > >>>>>> Best > >>>>>> Martin | mawiesne > >>>>>> -- > >>>>>> [1] https://lists.apache.org/thread/nvzl4g2b6rc149nf54xpnorjso5h0mlp > >> <https://lists.apache.org/thread/nvzl4g2b6rc149nf54xpnorjso5h0mlp> > >>>>> > >>> > >>> Disclaimer > >>> > >>> The information contained in this communication from the sender is > >> confidential. It is intended solely for use by the recipient and others > >> authorized to receive it. If you are not the recipient, you are hereby > >> notified that any disclosure, copying, distribution or taking action in > >> relation of the contents of this information is strictly prohibited and may > >> be unlawful. > >>> > >>> This email has been scanned for viruses and malware, and may have been > >> automatically archived by Mimecast, a leader in email security and cyber > >> resilience. Mimecast integrates email defenses with brand protection, > >> security awareness training, web security, compliance and other essential > >> capabilities. Mimecast helps protect large and small organizations from > >> malicious activity, human error and technology failure; and to lead the > >> movement toward building a more resilient world. To find out more, visit > >> our website. > >> > >> >
