[infinispan-dev] Grid Filesystem

[Marko Lukša]

I've just submitted a pull request with the last of my enhancements and 
fixes.

I have implemented everything we discussed in previous emails, added 
random seek (in ReadableGridFileChannel), cleaned up the code and added 
tests.

If there's anything else you'd like me to do, just ping me.

Regards,
Marko



On 4.1.2012 10:06, Bela Ban wrote:
>
> On 1/3/12 9:33 PM, Marko Lukša wrote:
>
>> Yes, since in GFS there's no concept of "working dir" as there is in
>> regular filesystems, the working/base dir should always be "/".
>> So, do you think a basePath constructor parameter would come in handy to
>> anyone? If so, I can implement it (instead of just making sure paths
>> with a leading / are equivalent to paths without it).
>
> I suggest whatever you do, it should have the same semantics as the
> corresponding java.io.* classes.
>
>
>> One more question: since GFS is still experimental, is there any reason
>> against me renaming fields&  variables to the project's code style
>> (currently, some vars are named like "default_chunk_size"); also there's
>> probably no reason against me making fields in GridOutputStream private?
> Go ahead; this is my K&R style (I refuse to adopt the var/field naming
> conventions of Java !) :-)
>
>