[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-promo
Subject:    Re: [kde-promo] nepomuk / baloo naming implications
From:       Vishesh Handa <me () vhanda ! in>
Date:       2014-02-19 10:34:23
Message-ID: 2454184.Qz5dH735hK () vlap
[Download RAW message or body]

On Wednesday, February 19, 2014 02:04:51 AM Aaron J. Seigo wrote:
> hi...
> 
> as the other thread is a bit spread out in terms of topoic, i thought i'd
> start afresh and try to collect the implications in one exhaustive list
> (it's long .. sorry :/)
> 
> == This is a journey ==
> 
> This will take a long time to communicate, no matter what happens: name
> change or not. I expect 6 months of consistent messaging is a bare minimum,
> and it will likely take longer. This is not a one-time communication
> question, but something that sets the tone for a 6 month trek, and which
> will easily have implications for the next 3+ years.
> 
> == Established investment nepomuk ==
> 
> While we've tried to keep Nepomuk out of the default UI as much as possible,
> it still shows up in places. It is also required for troubleshooting, and
> as such exists on our support wikis. We have recognized visual identity
> components (the nepomuk logo icon) that have been more or less visible in
> Plasma for some time.
> 

Says who? Established users know what the icon means because they are used to 
it not because the icon means anything. Try showing that icon to a random 
person and asking what they think it does.

Also, it has not been visible in Plasma. It has just been visible in the 
System Settings.

> Being able to attach the "bigger, better, faster" story of Baloo to Nepomuk
> is an easier story to tell than having to explain why Baloo replaces
> Nepomuk. It also allows us to preserve the above investments already made.
> 
> == Existing "common knowledge" of what Nepomuk is ==
> 
> The core Free software user community knows what Nepomuk is and does: file
> indexing and desktop search. This is not actually accurate, but it is the
> prevailing common understanding of it since that is what people have
> experienced with it. Usage by Akonadi or Plasma Active is not well known;
> some even confuse KRunner with Nepomuk since they use KRunner to search for
> things (!)
> 
> IOW, people have a vague understanding of what Nepomuk does. If the details
> under the hood change, it really doesn't matter. Risk of confusion because
> the implementation details change but the name stays the same, therefore,
> is low.
> 
> We benefit by not having to re-explain what it is Nepomuk did and Baloo does
> now. In fact, as Carl pointed out: it is really hard to introduce Baloo
> without talking about Nepomuk .. which says a lot.

Except that Baloo and Nepomuk do NOT do the same things. Lets not confuse PR 
with tech speak. In the current code based all the "relationships" part of 
Baloo have been throw away because there were no users. Tags are being stored 
in the xattr, and activities will follow suit.

Baloo is NOT Nepomuk. It attempts to solve certain similar problems, but those 
problems aren't endemic to the Nepomuk world. They are *very* general 
problems.

- Dealing with things in higher level such as People instead of Contacts - 
Everyone is doing this! Everyone.

- Searching and Indexing - Common problem

- Forming relationships to remember stuff - This person has this facebook 
account. Nepomuk may have allowed generalized versions of this coupled with a 
rich query language, but everyone needs this.

> 
> == API change ==
> 
> Some have noted that there has been significant API changes. This is true ..
> except for the widgets. Keeping the name makes that part of the story
> easier to communicate, so the question is whether it is significantly
> easier to communicate an API change with a new name.
> 
> If we look at things like Perl 6, Python 3 or Qt4 (or Qt2), or frameworks
> like node.js as it evolved: API changes are part of life. Developers choose
> a framework based on what it does ("does this open zip files? ‘cause I need
> to open zip files"), maturity, commitment by the team behind it and
> quality.
> 
> Given people know what Nepomuk does, keeping the name is a win there.
> 

Except that Baloo does not do the same things. Also from the time I've worked 
with Nepomuk, people equated Nepomuk with RDF.

> Maturity and commitment are implied by continuity. Changing names and saying
> "this is essentially a brand new project" doesn't help there, even if the
> new thing is awesome. Compare the struggles of Linux middleware that has
> changed names repeatedly to something like NetworkManager which has kept
> its name even though its API keeps changing from release to release rather
> significantly. Since it's the same name but just "keeps getting better" it
> avoids much of the controversy that other API-shifting areas of Linux
> middleware have endured (fairly or not).
> 
> The quality issue is the big remaining issue ... but not limited to API
> 
> == The Quality Hump ==
> 
> This is an issue from both the end user and developer perspective.
> 
> Nepomuk has a relatively poor reputation (fairly or not). Keeping the name
> "Nepomuk" therefore carries some of this negativity. In the "keep the name"
> situation we then have the job of making the case for why/how the current
> work has improved .. well .. anything.
> 
> If we change the name, we need to still explain why/how the current work has
> improved anything. We also need to explain why Nepomuk was entirely
> abandoned. Which also implies "why we stuck with it for so long even though
> it needs abandoning".
> 

We've explained in the first email announcing Baloo that Nepomuk internally was 
too complex and that it could not be optimized any further in the current 
scheme of things.

We don't need to explain Baloo to people. It's a search index, done.

> So changing the name doesn't save us much work at all; it may even create
> *more* work. The current article has an entire section called "Why change
> Nepomuk?" That entire section can pretty much be scrapped if the name is
> kept. It can be recycled into a "how this new version is better" and merged
> with the content currently under "About Baloo"
> 
> == Deprecation story ==
> 
> Saying "The Nepomuk project is officially deprecated" is pretty harsh
> sounding. It leaves an immediate question: 'what to move to if I'm using
> Nepomuk now?' On the other hand, people are quite used to old versions
> being deprecated (that's almost axiomatic). Keeping the name makes the
> migration path a lot more obvious while sounding a lot less like we're
> abandoning things and more like we're evolving things.
> 
> == Migration Story ==
> 
> Keeping the name also softens the migration story. Compare:
> 
> "Nepomuk and Baloo can coexist without issues. However, it is not optimal to
> run both of them on the same system. They both would be indexing files,
> emails and other data, duplicating functions, taxing the system
> unnecessarily, populating and synchronizing their databases."
> 
> with:
> 
> "Nepomuk 1 and 2 can coexist without issues, however it is not recommended
> to run both on the same system. Doing so would result in indexing all
> files, emails and other data twice, duplicating functions and taxing the
> system unnecessarily due to populating and synchronizing two separate
> databases."
> 
> or
> 
> "KDE Applications and Platform 4.13 will not ship with Nepomuk based search
> integrated. Nepomuk will still be available for distributions to ship as a
> separate component, but it is recommended that users who depend on
> applications requiring the old version of Nepomuk to hold off upgrading to
> Platform 4.13 until these applications have been ported. Linux distributions
> will most likely take care of this, ensuring that if any component needs
> Nepomuk it will be installed and otherwise, Baloo will be providing the
> search and storage capabilities."
> 

Mostly just search. but if you really want to mention "storage" one can. I 
would not.

> with
> 
> "KDE Platform 4.13 will ship with Nepomuk 2 integration only. Likewise, KDE
> Applications 4.13 has also been upgraded to use Nepomuk 2 exclusively. This
> covers the majority of Nepomuk applications most people use. However, those
> relying on applications which have not yet been ported to Nepomuk 2 are
> strongly recommended to stay with KDE Platform 4.12 until those applications
> are ported."
> 
> It sounds more like migration and upgrade than an awkward replacement.
> 
> 
> == Research participation story ==
> 
> Up until now we've been able to say "Yes, we have benefited from engaging
> productively in public research projects like Nepomuk in the EU." That has
> gotten us quite a bit of kudos over the years.
> 
> If we keep the name, we can demonstrate continuity from that investment and
> chart from KDE's past efforts in it to where we are now. Yes, it evolved,
> but it started with that research endeavor brought into KDE. If we toss the
> name, here's what we end up with something like this:
> 
> "We engaged in a EU research project called Nepomuk. It needed significant
> improvements to reach the level of performance we required, so eventeually
> we replaced it with something called 'Baloo' which has a higher performance
> file indexer as well as a more focused API better for what applications
> actually needed. Some of the people who worked on Nepomuk also wrote Baloo,
> so Nepomuk was a valuable part of the path to our current solution."
> 
> That's really, really weak for trying to present our research involvement
> and the value of tossing resources at KDE projects. If we keep the name,
> this same story becomes:
> 
> "We engaged in a EU research project called Nepomuk. Even after the research
> project ended, we continued to invest in it and evolve the framework. Today
> we have Nepomuk <N> which has a higher performance file indexer as well as
> a more focused API that is better suited to what desktop applications
> actually needed. In this manner, we took the research project and turned it
> into a production product."
> 
> It is really hard to make that same story when the names are changing,
> particularly in front of people who hand out research grants.
> 

It seems more hypocritical to keep the name and not use any of the 
technologies researched.

> == Possible post-baloo name churn ==
> 
> From the "pessimists" department: *if* Baloo were not to work out as we want
> and ends up being replaced / augmented with something else in the next
> years, we'll have this whole name change issue all over again .. and then
> it would be 3 name changes for the same thing.
> 

Ditto for everything else in life.

If something doesn't work you move on. Keeping names for the sake of PR is 
just confusing and counter-intuitive.

I personally want KDE to be associated with good quality and a good user 
experience independent of what is under the hood. This is IMHO what end-users 
care about. Not these squabbles over names.

> (Which reminds me of Kat and Tenor :)
> 
> This will lead to KDE looking like it can't figure out what it is doing ...
> sort of like the Linux middleware name churn which certainly hasn't helped
> their cause in getting people to adopt new iterations without screaming
> bloody murder (and forking in some cases)

If that ends with better quality at the end of the day, then so be it.

I'd just like to conclude by saying that I am NOT on board with calling Baloo, 
Nepomuk. I agree that there has been some activity on social media, but that 
will pass. 

Lets focus on the positive parts and make sure "Desktop Search" in KDE works 
very well. Code reviews + patches are most welcome.

-- 
Vishesh Handa

_______________________________________________
This message is from the kde-promo mailing list.

Visit https://mail.kde.org/mailman/listinfo/kde-promo to unsubscribe, set digest on \
or temporarily stop your subscription.


[prev in list] [next in list] [prev in thread] [next in thread]