Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
SemanticString
03-13-2018, 05:57 PM
Post: #11
RE: SemanticString
(03-13-2018 05:54 PM)lingu Wrote:  To be clear, we split format, not Blen. Blen can still be much larger than 4GB.

Fixed.
Find all posts by this user
Quote this message in a reply
04-18-2018, 04:34 PM (This post was last modified: 04-18-2018 04:38 PM by YU_Xinjie.)
Post: #12
RE: SemanticString
(03-12-2018 09:00 PM)lingu Wrote:  
(02-11-2018 12:35 PM)lingu Wrote:  int64 format; // format ID. 1: raw byte sequence.
int64 Blen; // size in bytes not including format and Blen

Driving example, with composite SSes, we want to know how many SSes there are in a composite SS (CSS). Blen gives the total Bize but not the SS count.

It is also a waste to use 64b for format.

I suggest we split format into 32b format followed by 32b sscnt which is the count of SemanticString structures in this CSS. For a non-composite SemanticString, sscnt is always 0.

RR xinjie

I find sscnt_ is defined in SemanticString.java while it is only used in CompositeSemanticString.java.
Logically, sscnt_ should be along with sss_[], since sscnt_ means the element number of sss_[].

I suggest to move sscnt_ from SemanticString.java into CompositeSemanticString.java to achieve better encapsulation.

RR lingu
Find all posts by this user
Quote this message in a reply
04-18-2018, 09:47 PM
Post: #13
RE: SemanticString
(04-18-2018 04:34 PM)YU_Xinjie Wrote:  
(03-12-2018 09:00 PM)lingu Wrote:  
(02-11-2018 12:35 PM)lingu Wrote:  int64 format; // format ID. 1: raw byte sequence.
int64 Blen; // size in bytes not including format and Blen

Driving example, with composite SSes, we want to know how many SSes there are in a composite SS (CSS). Blen gives the total Bize but not the SS count.

It is also a waste to use 64b for format.

I suggest we split format into 32b format followed by 32b sscnt which is the count of SemanticString structures in this CSS. For a non-composite SemanticString, sscnt is always 0.

RR xinjie

I find sscnt_ is defined in SemanticString.java while it is only used in CompositeSemanticString.java.
Logically, sscnt_ should be along with sss_[], since sscnt_ means the element number of sss_[].

I suggest to move sscnt_ from SemanticString.java into CompositeSemanticString.java to achieve better encapsulation.

RR lingu

Dont make the change before the test cases are stable.

Be aware that SemanticString present itself in the FS (both Linux local FS and GFAT) as a file. It is easy to remove sscnt_ from SemanticString. However, this makes it more complex to parse a SemanticString from the file data -- it can be a SemanticString or CompositeSemanticString, and based on which it is, the offsets may be different. This would create a couple of things that are not difficult to solve but cost time. We dont want to spend our life on that, particularly when we are disappointing customers.

If you are sure you want to remove that, please do it after the tests are stable. I feel this is textbook correct but in practice questionable.
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump: