lucenenet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Timothy Januario" <>
Subject RE: [jira] Commented: (LUCENENET-42) Hits class should have a friendlier interface
Date Mon, 29 Dec 2008 21:40:48 GMT
I agree that Lucene.Net would benefit from taking advantage of several
of the interfaces available in the .NET framework.  My intent in my
statement was certainly not to insult.  However, code divergence should
be kept to a minimum in order to allow for quick grepping of changes
from Lucene's Java code.  If the code changes too much between
implementations, it will become very difficult to find places where
changes are required.  My concern with the open issues is that they
should be resolved as quickly as possible into some status other than
OPEN (my suggestion was BY DESIGN but UNDER CONSIDERATION would be
equally acceptable) so that they don't cloud the environment from issues
that need resolution.


-----Original Message-----
From: Granroth, Neal V. [] 
Sent: Monday, December 29, 2008 2:02 PM
Subject: RE: [jira] Commented: (LUCENENET-42) Hits class should have a
friendlier interface

We should be careful not to quickly close issues using a reason of "By
Design".  Doing so usually only serves to insult the issue submitter and
does not indicate why the requested change cannot be made.

While keeping close to the original Java code is a good general
philosophy, we should make use of the "better", more useful, interfaces
available in .NET when we can.  Making Lucene.NET as .NET friendly as
possible and not forcing people who may be unfamiliar with Java to use
its sometimes awkward paradigms will help more developers adopt
Lucene.NET and ensure the project's survival.

For those who may not be aware, there was at least one JIRA issue about
problems in earlier versions of Lucene.NET with partially implemented or
incorrectly implemented IEnumerable interfaces.  In these situations
.NET developers could write clearly valid code that would unexpectedly
fail when run due to differences between the Java and .NET paradigms.
One solution in these situations is to eliminate the IEnumerable
interface and force the developer to use the Java-style method call.
Not a .NET friendly solution, but it works.

Perhaps we should consider a set of wrapper classes (adapter pattern) to
isolate any Lucene.NET augmentation from the core Lucene code?

-- Neal

-----Original Message-----
From: Tim Januario (JIRA) []
Sent: Wednesday, December 24, 2008 10:01 AM
Subject: [jira] Commented: (LUCENENET-42) Hits class should have a
friendlier interface

09#action_12659109 ]

Tim Januario commented on LUCENENET-42:

Copied from my comments on LUCENENET-43:
I agree with you George on this subject. The philosophy of the project
is to make the port as close as possible to the java source which has
huge benefits moving forward as new releases are made and need porting.
Once divergences are made, it will not be possible to simply compare two
branches of the java code in order to identify the changes that need to
be made to the Lucene.Net code. In keeping with this philosophy, it
would be beneficial to those of us who are watching these threads to
close the issues as "By Design" in order to not cause clutter.

> Hits class should have a friendlier interface
> ---------------------------------------------
>                 Key: LUCENENET-42
>                 URL:
>             Project: Lucene.Net
>          Issue Type: Improvement
>            Reporter: Simone Busoli
>            Priority: Trivial
> Retrieving the information about each hit is not very straightforward.
It is actually implemented as a method - Hits.Doc(i) - but .NET provides
indexers which are much more usable, that is, Hits[i]. Furthermore it
should be made easier to iterate through hits, implementing the
IEnumerable interface.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message