Thank you Maxim. There is CASSANDRA-18717, I guess that patch should go there. Keeping the task or not, the fix of the docs should go in anyway IMHO. I will not be available the next few days, but I can help with reviews when I am back.
On Thu, 3 Aug 2023 at 17:44, Maxim Muzafarov <mmu...@apache.org> wrote: > Yes, I agree. The javadoc task should be part of our CI if we decide > to keep it, to keep it buildable at all times. > > > BTW, I have managed to fix all the javadoc errors. > I have tested the task for both jdk11 and jdk17. > > Changes are here: > > https://github.com/apache/cassandra/compare/trunk...Mmuzaf:cassandra:javadoc_build > > On Thu, 3 Aug 2023 at 21:20, Ekaterina Dimitrova <e.dimitr...@gmail.com> > wrote: > > > > Thank you Maxim, > > > > “ > > > > From my point of > > view, the problem is that the javadoc task is not given the attention > > it deserves. The failonerror is currently 'false' and the task itself > > is not a part of any build and/or release processes, correct me if I'm > > wrong. > > > > So, > > 1. Fix warnings/errors; > > 2. Make the javadoc task part of the build (e.g. put it under > > 'artifacts'), or make it part of the release process that is regularly > > checked on the CI; > > 3. Publish/deploy the javadoc htmls for release in the special > > directory of the cassandra website to give them a chance of being > > indexed;“ > > > > This is aligned with what I saw and the two options mentioned at the > beginning - if we decide to keep it we should fix things and add the task > to CI, if we don’t because no one wants the html pages - then better to > remove it this ant task. > > On your comment about 100 errors - it seems they are more. There is a > cap of 100 but when you fix them, more errors appear. > > Further discussion can be found at CASSANDRA-17687 > > > > On Thu, 3 Aug 2023 at 14:21, Maxim Muzafarov <mmu...@apache.org> wrote: > >> > >> Personally, I find javadocs quite useful, especially when htmls are > >> indexed by search engines, which in turn increases the chances of > >> finding the right answer faster (I have seen a lot of useful javadocs > >> in the source code). > >> > >> I have done a quick build of the javadocs: > >> > >> [javadoc] Building index for all the packages and classes... > >> [javadoc] Building index for all classes... > >> [javadoc] Building index for all classes... > >> [javadoc] 100 errors > >> [javadoc] 100 warnings > >> > >> 100 errors is no big deal and can be easily fixed. From my point of > >> view, the problem is that the javadoc task is not given the attention > >> it deserves. The failonerror is currently 'false' and the task itself > >> is not a part of any build and/or release processes, correct me if I'm > >> wrong. > >> > >> So, > >> 1. Fix warnings/errors; > >> 2. Make the javadoc task part of the build (e.g. put it under > >> 'artifacts'), or make it part of the release process that is regularly > >> checked on the CI; > >> 3. Publish/deploy the javadoc htmls for release in the special > >> directory of the cassandra website to give them a chance of being > >> indexed; > >> > >> On Thu, 3 Aug 2023 at 17:11, Jeremiah Jordan <jeremiah.jor...@gmail.com> > wrote: > >> > > >> > I don’t think anyone wants to remove the javadocs. This thread is > about removing the broken ant task which generates html files from them. > >> > > >> > +1 from me on removing the ant task. If someone feels the task is > useful they can always implement one that does not crash and add it back. > >> > > >> > -Jeremiah > >> > > >> > On Aug 3, 2023 at 9:59:55 AM, "Claude Warren, Jr via dev" < > dev@cassandra.apache.org> wrote: > >> >> > >> >> I think that we can get more developers interested if there are > available javadocs. While many of the core classes are not going to be > touched by someone just starting, being able to understand what the > external touch points are and how they interact with other bits of the > system can be invaluable, particularly when you don't have the entire code > base in front of you. > >> >> > >> >> For example, I just wrote a tool that explores the distribution of > keys across multiple sstables, I needed some of the tools classes but not > much more. Javadocs would have made that easy if I did not have the source > code in front of me. > >> >> > >> >> I am -1 on removing the javadocs. > >> >> > >> >> On Thu, Aug 3, 2023 at 4:35 AM Josh McKenzie <jmcken...@apache.org> > wrote: > >> >>> > >> >>> If anything, the codebase could use a little more > package/class/method markup in some places > >> >>> > >> >>> I am impressed with how diplomatic and generous you're being here > Derek. :D > >> >>> > >> >>> On Wed, Aug 2, 2023, at 5:46 PM, Miklosovic, Stefan wrote: > >> >>> > >> >>> That is a good idea. I would like to have Javadocs valid when going > through them in IDE. To enforce it, we would have to fix it first. If we > find a way how to validate Javadocs without actually rendering them, that > would be cool. > >> >>> > >> >>> There is a lot of legacy and rewriting of some custom-crafted > formatting of some comments might be quite a tedious task to do if it is > required to have them valid. I am in general for valid documentation and > even enforcing it but what to do with what is already there ... > >> >>> > >> >>> ________________________________________ > >> >>> From: Jacek Lewandowski <lewandowski.ja...@gmail.com> > >> >>> Sent: Wednesday, August 2, 2023 23:38 > >> >>> To: dev@cassandra.apache.org > >> >>> Subject: Re: [DISCUSSION] Shall we remove ant javadoc task? > >> >>> > >> >>> NetApp Security WARNING: This is an external email. Do not click > links or open attachments unless you recognize the sender and know the > content is safe. > >> >>> > >> >>> > >> >>> > >> >>> With or without outputting JavaDoc to HTML, there are some errors > which we should maybe fix. We want to keep the documentation, but there can > be syntax errors which may prevent IDE generating a proper preview. So, the > question is - should we validate the JavaDoc comments as a precommit task? > Can it be done without actually generating HTML output? > >> >>> > >> >>> Thanks, > >> >>> Jacek > >> >>> > >> >>> śr., 2 sie 2023, 22:24 użytkownik Derek Chen-Becker < > de...@chen-becker.org<mailto:de...@chen-becker.org>> napisał: > >> >>> Oh, whoops, I guess I'm the only one that thinks Javadoc is just > the tool and/or it's output (not the markup itself) :P If anything, the > codebase could use a little more package/class/method markup in some > places, so I'm definitely only in favor of getting rid of the ant task. I > should amend my statement to be "...I suspect most people are not opening > their browsers and looking at Javadoc..." :) > >> >>> > >> >>> Cheers, > >> >>> > >> >>> Derek > >> >>> > >> >>> > >> >>> > >> >>> On Wed, Aug 2, 2023, 1:30 PM Josh McKenzie <jmcken...@apache.org > <mailto:jmcken...@apache.org>> wrote: > >> >>> most people are not looking at Javadoc when working on the codebase. > >> >>> I definitely use it extensively inside the IDE. But never as a > compiled set of external docs. > >> >>> > >> >>> Which is to say, I'm +1 on removing the target and I'd ask everyone > to keep javadoccing your classes and methods where things are non-obvious > or there's a logical coupling with something else in the system. :) > >> >>> > >> >>> On Wed, Aug 2, 2023, at 2:08 PM, Derek Chen-Becker wrote: > >> >>> +1. If a need comes up for Javadoc we can fix it at that point, but > I suspect most people are not looking at Javadoc when working on the > codebase. > >> >>> > >> >>> Cheers, > >> >>> > >> >>> Derek > >> >>> > >> >>> On Wed, Aug 2, 2023 at 11:11 AM Brandon Williams <dri...@gmail.com > <mailto:dri...@gmail.com>> wrote: > >> >>> I don't think even if it works anyone is going to use the output, so > >> >>> I'm good with removal. > >> >>> > >> >>> Kind Regards, > >> >>> Brandon > >> >>> > >> >>> On Wed, Aug 2, 2023 at 11:50 AM Ekaterina Dimitrova > >> >>> <e.dimitr...@gmail.com<mailto:e.dimitr...@gmail.com>> wrote: > >> >>> > > >> >>> > Hi everyone, > >> >>> > We were looking into a user report around our ant javadoc task > recently. > >> >>> > That made us realize it is not run in CI; it finishes > successfully even if there are hundreds of errors, some potentially > breaking doc pages. > >> >>> > > >> >>> > There was a ticket discussion where a few community members > mentioned that this task was probably unnecessary. Can we remove it, or > shall we fix it? > >> >>> > > >> >>> > Best regards, > >> >>> > Ekaterina > >> >>> > >> >>> > >> >>> -- > >> >>> +---------------------------------------------------------------+ > >> >>> | Derek Chen-Becker | > >> >>> | GPG Key available at https://keybase.io/dchenbecker< > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fkeybase.io%2Fdchenbecker&data=05%7C01%7CStefan.Miklosovic%40netapp.com%7C7ca04f0f58764996ab1e08db93a0de2a%7C4b0911a0929b4715944bc03745165b3a%7C0%7C0%7C638266091373361824%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=n%2BrDfikzzoQG%2Fg%2BRvNqEEE6vHP8ZmY1skeosesLK9v0%3D&reserved=0> > and | > >> >>> | https://pgp.mit.edu/pks/lookup?search=derek%40chen-becker.org< > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpgp.mit.edu%2Fpks%2Flookup%3Fsearch%3Dderek%2540chen-becker.org&data=05%7C01%7CStefan.Miklosovic%40netapp.com%7C7ca04f0f58764996ab1e08db93a0de2a%7C4b0911a0929b4715944bc03745165b3a%7C0%7C0%7C638266091373518054%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Tnu5cIoIFZGqhaqOjCjW8yK%2BDTT2%2B0ifvFNs1pJO93s%3D&reserved=0> > | > >> >>> | Fngrprnt: EB8A 6480 F0A3 C8EB C1E7 7F42 AFC5 AFEE 96E4 6ACC | > >> >>> +---------------------------------------------------------------+ > >> >>> > >> >>> > >> >>> > >> >>> >
