lucenenet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shad Storhaug <s...@shadstorhaug.com>
Subject RE: [Vote] Apache Lucene.Net 4.8.0-beta00001
Date Tue, 09 May 2017 10:49:03 GMT
Itamar,

Thanks for your input. You make a compelling argument.

Since the vote has passed and 4.8.0-beta00001 is already burnt (it exists in some people's
NuGet cache and if we re-use it we can't be sure if they are testing the right copy), let's
compromise and do both. Releasing now will do some damage control on the bootleg (which seriously
needs to be made clear that it is not official and not production-ready) and ensures we reserve
all of our NuGet package IDs. Starting a vote on 4.8.0-beta00002 now will ensure the bug will
be fixed within the same 72 hour timeframe.

We should be able to determine by the nature of the bug reports if they are definitely not
related to this and be able to fix those. Issues we are unsure about we can ask the users
whether they still experience them after upgrading to 4.8.0-beta00002 and close if that patch
fixes the issue(s). 

Peter has provided a workaround for the bug, which we can put into the release notes on NuGet.

We can hold off any official announcement until after 4.8.0-beta00002 is released. 

Thoughts?

Shad Storhaug (NightOwl888)


-----Original Message-----
From: itamar.synhershko@gmail.com [mailto:itamar.synhershko@gmail.com] On Behalf Of Itamar
Syn-Hershko
Sent: Tuesday, May 9, 2017 4:34 PM
To: dev@lucenenet.apache.org
Subject: Re: [Vote] Apache Lucene.Net 4.8.0-beta00001

This is quite a severe bug, and actually can cause index corruption. It can potentially also
crash the application - some tests have been indeed failing with an exception being thrown
due to access attempt of non-existing files. It is also probably going to fix quite a handful
of those flakey tests (which will take a while to notice). If it wasn't that critical, I would
have voted +1. In fact, I will probably cast an automatic
+1 on the next vote.

Tagging a version as official Beta, and having an announcement around it is bigger than just
having the bits around (which we had as a while).
Releasing a cleaner version will allow us to work on actual real bugs as they will be reported,
instead of potentially responding to bug reports on something we know is already fixed even
before we released. This is a better way of "collecting information" as you said.

The compilation issues Simon has identified are important to fix (I had some myself) but do
not constitute as critical IMO.

We can start another vote now, and like I said 72 hours delay is not a big deal.

--

Itamar Syn-Hershko
Freelance Developer & Consultant
Elasticsearch Partner
Microsoft MVP | Lucene.NET PMC
http://code972.com | @synhershko <https://twitter.com/synhershko> http://BigDataBoutique.co.il/

On Tue, May 9, 2017 at 12:19 PM, Shad Storhaug <shad@shadstorhaug.com>
wrote:

> Stefan,
>
> > If you run into it, will it make your application crash or will it
> destroy the index?
>
> It causes a crash under highly concurrent scenarios, and will most 
> likely affect all of the file-system directories. It does not affect 
> the index, otherwise some of the index tests would have detected it. 
> Peter van Ginkel (the user who discovered it) has been kind enough to 
> contribute a test that fails most of the time if the concurrency bug 
> exists, but before this none of our tests have been able to detect it. 
> Peter also has been able to work around this bug, and I have asked him to post the workaround
at:
> https://github.com/apache/lucenenet/pull/205
>
> It is a severe bug. Is it our most severe bug? Maybe. Is it severe 
> enough to destroy our reputation? Being that there is a bootleg copy 
> out there that is already doing just that (that is versioned as 
> production-ready and already has this bug), I would say we are better 
> off releasing with the bug than not. If we didn't have that issue to 
> contend with, I would agree with Itamar that we should re-roll the release.
>
> Thanks,
> Shad Storhaug (NightOwl888)
>
>
>
>
> -----Original Message-----
> From: Stefan Bodewig [mailto:bodewig@apache.org]
> Sent: Tuesday, May 9, 2017 11:01 AM
> To: dev@lucenenet.apache.org
> Subject: Re: [Vote] Apache Lucene.Net 4.8.0-beta00001
>
> On 2017-05-09, Shad Storhaug wrote:
>
> > So technically the vote passes. However, I will give it some more 
> > time
> in case anyone else wants to weigh in on whether the issues we have 
> are significant enough to reset the release. Presscott, Stefan, Simon, WDYT?
>
> As you may know I'm not a user of Lucene.Net myself, so take my 
> opinion with a grain of salt.
>
> I'm not sure about the impact of the bug. If you run into it, will it 
> make your application crash or wil it destroy the index? In the later 
> case I'd say we should re-roll the release. Otherwise we should 
> publish the release, fix the bug and plan for a second beta soon.
>
> Stefan
>
Mime
View raw message