I misunderstood your response Larry. I apologize about that. I suppose I can
always test the 3 different options for my implementation, but that would be
a lot of work. :] I thought some of the developers would have an idea on
which technique might be the fastest.
If we decide to move to a JDK that supports generics, then I will probably
tweak the implementation of the Queue interface as Michael suggested.
You wrote: "When I said I didn't understand what problem you are trying to
solve, I wasn't talking about queues and such. I was asking (big picture)
what OpenJump deficiency are you trying to remedy?"
I am trying to fix the problem OpenJUMP has with reading very large data
sets. This may not be something you run into a lot with facilities
management, but in my line of work it happens frequently. For example,
OpenJUMP chokes when I try to read in the parcels layer for my county, and
I'm running a CAD workstation. :] The FeatureCache will implement
"on-demand" features by managing a cache of features stored on the hard disk
in binary files. The "buffer" is meant to improve performance of the cache
by holding a small subset of the cache in memory.
That was the long answer. The short answer to your question is that I'm
trying to overcome the RAM limitation on the size of spatial data sets that
OpenJUMP can import and work with.
The Sunburned Surveyor
On 4/13/07, Larry Becker <[EMAIL PROTECTED]> wrote:
The best advice I can give on optimization is never do it until you have
working code. I have been working on JUMP for three years and have only
started looking at optimization the last month or so.
See http://www.extremeprogramming.org/rules/optimize.html
When I said I didn't understand what problem you are trying to solve, I
wasn't talking about queues and such. I was asking (big picture) what
OpenJump deficiency are you trying to remedy?
Larry
On 4/13/07, Sunburned Surveyor <[EMAIL PROTECTED]> wrote:
>
> Michael and Larry,
>
> Thank you for the responses. Please see my comments below.
>
> Michael wrote: "Did you have a look to java 5 documentation ? : you'll
> find interesting
> information in Queue interface and LinkedList implementation."
>
> Yes, I took a look at the Queue interface. It had the First-In-First-Out
> behavior, but I didn't see a way to limit the growth of the content, which I
> would need to do for my purposes. If we decide to make the JUMP to a newer
> JDK version that supports generics I might take a look at this.
>
> Michael wrote: "I think there are also many open-source projects related
> to Cache
> management."
>
> Could you toss me a name or two? :] Remember that this question was
> really about just the buffer, which is one part of the whole FeatureCache.
>
> Larry wrote: "Regarding specific suggestions, I'm afraid I haven't yet
> understood exactly what problem you are trying to solve."
>
> I'm trying to find the most efficient and fast First-In-First-Out
> collection for in-memory representation of features. I was concerned the
> cost of object casting, since Java 1.4.2 doesn't support generics. Java
> 1.4.2 also does not contain a collection that implements the
> "growth-to-a-limit" behavior that I need.
>
> I will visit the JDK Version topic in a new thread, because I feel it is
> an important one.
>
> The Sunburned Surveyor
>
>
> On 4/13/07, Larry Becker < [EMAIL PROTECTED]> wrote:
>
> > Hi Sunburned,
> >
> > Michaël is correct. It is time to embrace the Java 5 enhancements.
> > It may even be time to start considering Java 6. My testing shows no
> > problems with compatibility.
> >
> > Regarding specific suggestions, I'm afraid I haven't yet understood
> > exactly what problem you are trying to solve.
> >
> > regards,
> > Larry
> >
> > On 4/13/07, Michaël Michaud <[EMAIL PROTECTED] > wrote:
> > >
> > > Hi,
> > >
> > > Did you have a look to java 5 documentation ? : you'll find
> > > interesting
> > > information in Queue interface and LinkedList implementation.
> > > I think there are also many open-source projects related to Cache
> > > management.
> > > With java 5 generics cast is no more necessary.
> > > My advices :
> > > - read the javadoc first
> > > - use java 5 (this subject has not been discussed for a long time,
> > > but
> > > my personnal feeling is that it is now time to use java 5, specially
> > > for
> > > a new important project / feature)
> > > - do-it yourself is good to learn, but for general problems and
> > > performance issues, existing libs is often a better choice :-)
> > >
> > > My two cents
> > >
> > > Michaël
> > >
> > >
> > > Sunburned Surveyor a écrit :
> > >
> > > > I've been doing some more work on my FeatureCache implementation.
> > > I am
> > > > currently designing a "buffer" that will hold a set number of
> > > features
> > > > from the feature cache in memory. This will increase performance
> > > when
> > > > a user is working with the same small group of features. The
> > > maximum
> > > > number of features in the buffer will be set by the user and can
> > > be
> > > > based on RAM of the computer running OpenJUMP and the user's need
> > > for
> > > > speed. (A default maximum feature count will be provided.)
> > > >
> > > > I need some help from our more experienced Java developers. Larry
> > > > seems to have a nack for performance issues, so perhaps he will
> > > have
> > > > some advice. Any suggestion are welocome. :]
> > > >
> > > > I'm trying to figure what type of collection/container to use for
> > > the
> > > > buffer. I took a look at the existing Java Collection
> > > implementations,
> > > > and I don't see one that will work out of the box. I need a
> > > > First-In-First-Out collection whose growth I can limit. I think I
> > > have
> > > > three choices for the buffer's container:
> > > >
> > > > [1] Extend and modify an existing Collection implementation.
> > > > [2] Write my own implementation of the Collection interface that
> > > > behaves the way I need it to.
> > > > [3] Write an implemenation that uses an array internally, works
> > > > directly with objects that implement the Feature interface, but
> > > that
> > > > does not implement Collection.
> > > >
> > > > Option 1 is probably the easiest, but I don't think it will be the
> > >
> > > > most efficient or fast. Option #2 is great from a reusability
> > > > standpoint, but it is a lot more work. I think Option 3 will be
> > > the
> > > > fastest and most efficient, becuase I won't have to make object
> > > casts,
> > > > but it will be more work than Option 1.
> > > >
> > > > How much speed will I gain if I avoid the Collection interface and
> > > the
> > > > resulting casts from Object to Feature? Do you think Option 2 will
> > > be
> > > > significantly faster?
> > > >
> > > > I think speed will be critical for this part of the FeatureCache
> > > > implementation. The only part more critical from a speed point of
> > > view
> > > > will be the binary format reader/writer.
> > > >
> > > > Thanks,
> > > >
> > > > The Sunburned Surveyor
> > > >
> > > >
> > >
> > > >------------------------------------------------------------------------
> > > >
> > > >-------------------------------------------------------------------------
> > >
> > > >Take Surveys. Earn Cash. Influence the Future of IT
> > > >Join SourceForge.net's Techsay panel and you'll get the chance to
> > > share your
> > > >opinions on IT & business topics through brief surveys-and earn
> > > cash
> > > >http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> > >
> > > >
> > > >------------------------------------------------------------------------
> > >
> > > >
> > > >_______________________________________________
> > > >Jump-pilot-devel mailing list
> > > > [EMAIL PROTECTED]
> > > > https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
> > > >
> > > >
> > >
> > >
> > >
> > > -------------------------------------------------------------------------
> > > Take Surveys. Earn Cash. Influence the Future of IT
> > > Join SourceForge.net's Techsay panel and you'll get the chance to
> > > share your
> > > opinions on IT & business topics through brief surveys-and earn cash
> > >
> > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> > > _______________________________________________
> > > Jump-pilot-devel mailing list
> > > [EMAIL PROTECTED]
> > > https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
> > >
> >
> >
> >
> > --
> > http://amusingprogrammer.blogspot.com/
> >
> > -------------------------------------------------------------------------
> > Take Surveys. Earn Cash. Influence the Future of IT
> > Join SourceForge.net's Techsay panel and you'll get the chance to
> > share your
> > opinions on IT & business topics through brief surveys-and earn cash
> >
> > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> > _______________________________________________
> > Jump-pilot-devel mailing list
> > [EMAIL PROTECTED]
> > https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
> >
> >
>
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share
> your
> opinions on IT & business topics through brief surveys-and earn cash
>
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Jump-pilot-devel mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
>
>
--
http://amusingprogrammer.blogspot.com/
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share
your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Jump-pilot-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Jump-pilot-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
