1. FW: Re[2]: Suggestion for EDS
- Posted by Henri.Goffin at sbs.be
May 25, 2001
I got the same 3meg file but, stangely enough, if you replace:
isi &= 254
with
isi &= rand(255)
you got a file that is 1.1MB which is reasonably proportionate.
H. Goffin
> -----Original Message-----
> From: aku at inbox.as [SMTP:aku at inbox.as]
> Sent: Friday, May 25, 2001 16:20
> To: EUforum
> Subject: Re[2]: Suggestion for EDS
>
>
>
> R> Aku writes:
> >> 1. if user stores a sequence, make EDS can check if it is
> >> a string (all bytes 0-255). If yes, store as string (just use 1
> >> byte per atom). If not, store as usual. If this is make EDS slower
> >> make function enable_string() and disable_string() for example.
> >> So I can store contents of a 1 MB file without having a 4 MB .edb
> >> file.
>
> R> EDS only uses one byte per character for strings
> R> (except for extremely rare cases where 2 bytes might be used).
> R> There's also a couple of bytes of overhead.
>
> Sorry I dont know that.
> but if:
>
> include database.e
>
> procedure x(object notused)
> end procedure
>
> x(db_create("tes",DB_LOCK_NO))
> x(db_open("tes",DB_LOCK_NO))
> x(db_create_table("file"))
> x(db_select_table("file"))
>
> object isi
> isi = {}
> for i=1 to 1000000 do
> isi &= 254 -- large atom
> end for
>
> x(db_insert("file",isi))
> db_close()
>
>
> the database file size is 3MB
> (3 bytes per atom)
>
> R> See compress()/decompress() in database.e
>
> >> 2. can make autonumber especially for record keys.
> >> and add indexing in same file (.edb) for record data.
>
> R> I'll think about it.
<snip>
>
