incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mehdi El Fadil <mehdi.elfa...@mango-is.com>
Subject Re: 2011 Apache CouchDB Community Survey...
Date Wed, 30 Nov 2011 15:16:16 GMT
On Wed, Nov 16, 2011 at 3:01 PM, Jan Lehnardt <jan@apache.org> wrote:

> I'd be interested to hear where exactly we fail to look after users. Note
> that this is a voluntary effort and we can't possibly make everybody happy.
> That said, we're trying and I'd be happy to learn where we can improve.
>

Hi Jan,

Here is the promised feedback about my experience, I hope it will be useful:

Before starting, I should tell that overall my experience with couchdb is
positive. My goal here is to provide a hopefully constructive feedback,
about some weaknesses I have felt about couchdb.

Doubts I have are in the line of some opinions which have already been
expressed here: http://news.ycombinator.com/item?id=3180397.

Alex Popescu comments about it: "If only *they* would have listened to what
I’ve been saying all this time. " (
http://nosql.mypopescu.com/post/12237740756/what-happened-to-couchdbs-popularity
)

The bottom line of my opinion is that you guys are too tech-focused, and
are ignoring what couchdb users need. By users I mean those who are not
that technical, not into couchdb internals, those who might decide in the
companies whether this technology will be chosen for the next project (and
these guys are not always technical).

Here is more detail, where I think there could be improvement (in no
particular order):

   - *Involving your 'customers' in the selection of features to add*. Yes,
   users have the possibility to enter features requests into Jira, but this
   requires a proactive effort from their part, + registration. What about
   polling the community from time to time via a survey? or just analyzing the
   recurring questions inside users mailing list? My feeling right now is that
   features are proposed / decided by technical people very close to the
   development of couchdb. It is all about technical optimizations, and not
   about functional optimizations. I am pretty sure, if you ask users, they
   will come up with ideas you did not think of and might think are good.
   - *Roadmap communication*: we -users, non couchdb developers - do not
   know where couchdb is heading. The content of Jira is too technical and
   there are too many info in there. Having a single page summarizing the next
   big changes would definitely help.
   - *Better promotion of the couchdb ecosystem*: I understand that you -
   in the context of Apache couchdb project - might want to keep focusing on
   the couchdb engine, and that you consider that side utilities like Futon or
   the jQuery library do not deserve more effort to put on. However, these
   tools to interact with couchdb are essential to the rest of us. It might be
   worth promoting the existing ecosystem around couchdb, like maintaining a
   directory of tools to import/export/administer/query/index/deploy/scale
   couchdb. Many things exist around, but not that easy to find, and it is not
   clear about maturity of each, and whether they are still maintained. Yes,
   community, users like me, should participate in such activities by
   maintaining the wiki. Send us a reminder via users mailing list from time
   to time maybe?
   - *Documentation*: There is a documentation problem in couchdb.
   Documents exist, some things are very well documented, some other not.
   Things got outdated with newer versions of couchdb but are not updated (eg.
   new security features not mentioned in old documents).
   - The following might hurt, sorry about that: the couchdb definitive
      guide is not as well written as advertised. If you are looking for one
      specific info on a topic and open the corresponding chapter, it might be
      ok. If you get started with couchdb and take this book to read for a
      week-end, then you notice the problems. Some ideas are repeated so many
      times chapter after chapter. Also, versions 1 and draft co-exist, with no
      indication whatsoever of their differences. Content seems to be the same.
      In case there is no update going on, then just hide this draft version.
      - Wiki could definitely have more content (eg. jquery.couch.js), be
      more updated and show more examples.
      - damn what do you have against jquery.couch.js? Documentation moved
      from couch.io to couchone then to couchbase... and then it somehow
      disappeared from there. and damn in 2011 no search features in
      couchbase.com blog? There are Dale Harvey's, Lena Lena's and Bradley
      Holt's, but none of them is really complete (or at least few
Months ago it
      was like that)
      - couchdb businesses down here: couchbase, cloudant... should
      definitely create a library with complete, up-to-date how-to
guides on the
      topics people keep asking all the time.
      - *Support*: some technical questions in the users mailing list
   receive no answer. Not often fortunately, but when it happens it's too bad.
   For couchdb volunteers I understand they work on a voluntary basis. This
   comment is therefore more directed to the ones also working at couchbase,
   cloudant, or other businesses depending on couchdb, looks like their
   interest to insure good customer support via the mailing list also.
   - *Marketing couchdb* *and credibility towards companies*: I think you
   should improve the way how couchdb is "sold". Of course it all depends on
   what your goal is, having couchdb as an open source project. Is it: we know
   what we are doing, we do a great tool, and we let the others find out by
   themselves that it is", or is it "We create couchdb to serve a purpose,
   listen to the users in order to have the features matching to their
   requirements, and show them that the tool fits their needs not any need, of
   course you keep focus on something)". At some point there was confusion
   between the various distributions of couchdb, I think you guys have noticed
   that. I have been evaluating couchdb as a solution for my current startup
   project. I have had serious doubts about whether I could find easily
   skilled couchdb developers, and I am still not convinced that it is the
   case. I have asked the question once and was told 'no no, there are plenty
   of couchdb developers'. I don't need to be told that, I need you to show it
   to me. I have been looking for someone able to develop a multidimensional
   view engine on the side of the existing view mechanism, result was 404 -
   not found.

I hope these words will be helpful to you.

Cheers,

Mehdi.

P.S.: according to points of views couchdb might be in an identity crisis
right now. You guys still have potential, and being in the shadow of the
nosql hype can be an opportunity to evolve in a more innovative direction
than where most people are driving nosql, and create something (more)
different. keep up the good work.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message