1. FILEMAN.E
Well, sad to say I was ill all weekend, so FILEMAN.E remains unchanged. At
this point, I am ready to begin re-writing it as per many of the suggestions
proposed. The biggest change will be adding 1 extra byte to the length of
each record, to manage deleted records. I checked into putting an ASCII 26
into the header of a file, but the only program which could not detect that
there was more data in the file was the 'type' command in DOS, so I'm going
to leave that be for now.
I am adding versioning into the header, just in case. At this point, I
think that the file format is set in stone now, but those are famous last
words, so I'm using versioning anyway.
The next update is a minor revision, so I won't be doing anything fancy like
record-locking or indexing yet. This is mostly to lock in a file format
that everyone is happy with, so I can convert my personal programs to
FILEMAN, and write a few demos.
If anyone has any last minute thoughts or comments, please let me know, as
I should be converting the code this evening.
Thanks,
Brian Jackson
2. Re: FILEMAN.E
Suggestion
When you delete a record just overwrite a ~ (tilde) character
over the first position in the deleted record. To compact the file
all you have to do is find the first record in file that begins
with ~ (tilde) and over write it with the next record if that record
is deleted skip to the next record and etc. I hope you understand
what I mean. This would be done in a buffer in memory and
would be fairly fast. You could also use a filter that reads
in a file and outputs a file while skiping the ~ (tilde) records.
The advantage is speed, easy way to delete, and it doesn't cost a byte.
Hope you feel better soon
Bernie
a good record until all deleted records are removed. By the way
if
3. Re: FILEMAN.E
>When you delete a record just overwrite a ~ (tilde) character
>over the first position in the deleted record.
That was my initial thought as well, but once you overwrite a character
in the record, you have no way to recover it. It will be a big pain in
the neck to re-write FILEMAN this way, but it preserves the entire
record for recovery later on. And if anyone is deleting lots of records
from a file, it's probably going to get large enough that one extra byte
in every record will hardly be noticeable.
>To compact the file all you have to do is find the first record in file
>that begins with ~ (tilde) and over write it with the next record if
>that record is deleted skip to the next record and etc. I hope you
>understand what I mean. This would be done in a buffer in memory and
>would be fairly fast. You could also use a filter that reads
>in a file and outputs a file while skiping the ~ (tilde) records.
That's pretty close to what I had in mind. I won't overwrite deleted
records, but there will be a command line utility (REFORMAT.EX) to
purge all the deleted records from a file. Again, the advantage here
is to be able to recover a deleted record (and it keeps the most
recently written data in one spot at the end of the file). When the
file gets too large, reformatting it will permanently erase all the
deleted records and shrink the file size down.
>Hope you feel better soon
Thanks :)
Brian Jackson
4. Re: FILEMAN.E
I haven't look at Fileman.e but may I suggest you try to use xBase-like
commands. In case you didn't knew, xBase is the term used to describe dBase
like languages (like FoxPro, Clipper, Alaska, etc.) that use the .dbf file
format (with some extensions). Purging deleted records is usually done with
the pack command. If you want take a look at my dbf engine, it's available
at RDS's contribution page.
Regards,
Daniel Berstein
[daber at pair.com]
5. Re: FILEMAN.E
Hi,
I agree to Daniel Berstein. I'm a xBase programmer since 1991. I look
forward to using FILEMAN.E. I was really hoping for someone to
come up with routines that I can use to access Fox Pro 2 tables since
I don't have the time to do it myself. But FILEMAN.E would be just fine.
"I haven't look at Fileman.e but may I suggest you try to use xBase-like
commands. In case you didn't knew, xBase is the term used to describe
dBase like languages (like FoxPro, Clipper, Alaska, etc.) that use the .dbf file
format (with some extensions)..."
Daniel Berstein
[daber at pair.com]
Thanks to everyone who contributes to the list. You're all an inspiration
for a Euphoria beginner like me.
Regards,
Arlie Codina
web.master at flashmail.com
6. Re: FILEMAN.E
>[Arlie}I agree to Daniel Berstein. I'm a xBase programmer since 1991.
>I look forward to using FILEMAN.E. I was really hoping for someone to
>come up with routines that I can use to access Fox Pro 2 tables since
>I don't have the time to do it myself. But FILEMAN.E would be just fine.
>
>[Daniel]"I haven't look at Fileman.e but may I suggest you try to use
>xBase-like commands. In case you didn't knew, xBase is the term used to
>describe dBase like languages (like FoxPro, Clipper, Alaska, etc.) that
>use the .dbf file format (with some extensions)..."
Well, guys, sad to say that I have absolutely ZERO experience with xBase,
so I'd be out of my league trying to merge FILEMAN and xBase :( However,
I may in the future have the time to check out Daniel's xBase routines (I
think he has some on his site), and try to figure it out.
Thanks for the suggestions :)
Brian Jackson
7. Re: FILEMAN.E
"Well, guys, sad to say that I have absolutely ZERO experience with xBase,
so I'd be out of my league trying to merge FILEMAN and xBase :( However,
I may in the future have the time to check out Daniel's xBase routines (I
think he has some on his site), and try to figure it out. --Brian Jackson--"
I'm enclosing a text on common xbase commands. I hope this will help.
Regards,
Arlie Codina
web.master at flashmail.com
oOo
*** XBase Common Data Manipulation Commands ***
CREATE - create or modify a table
+-field name
| +-field type
| | +- example: 1234.67 <---width is 7
| | | +-------index file name
| | | |
Name Type Width Decimal Index
FIRSTNAME CHAR 15 0 0
APPEND BLANK - adds a blank record to the end of the table and
makes the blank record the current record.
SEEK - searches for the first record in an indexed table whose key
fields matches the specified expression or expression list.
Syntax
SEEK <exp> | <exp list>
<exp> - the expression to search for in an index for a DBF
table.
<exp list> - one or more expressions, separated by commas,
to search for in a simple or composite key index for
non-DBF tables.
REPLACE - replaces the contents of fields with data from expressions.
Syntax
REPLACE
<field 1> WITH <exp 1> [ADDITIVE]
[, <field 2> WITH <exp 2> [ADDITIVE]...]
[<scope>]
[FOR <condition 1>]
[WHILE <condition 2>]
[REINDEX]
<field> WITH <exp> - designates fields to be replaced by the value
of the specified expressions. Multiple fields of a record
may be changed by including additional <field n> WITH <exp n>
expressions, separated by commas.
ADDITIVE - adds text to the end of memo field text instead of replacing
existing text. You can use ADDITIVE only when the specified
field is a memo field in a DBF table.
<scope>
FOR <condition 1>
WHILE <condition 2>
The scope of the command. The default scope is NEXT 1, the current record
only.
REINDEX - specifies that all affected indexes are rebuilt once the
REPLACE
operation finishes. Without REINDEX, Visual dBASE updates all
open indexes after replacing each record in the scope. When replacing many
or all records in a table that has multiple open indexes,
REPLACE
executes faster with the REINDEX option.
SKIP - moves the record pointer in the current or specified work area.
Syntax
SKIP [<expN>] [IN <alias>]
<expN> - the number of records Visual dBASE moves the record pointer
forward
or backward in the table open in the current or specified work
area.
If <expN> evaluates to a negative number, the record pointer
moves
backward. SKIP with no <expN> argument moves the record pointer
forward
one record.
IN <alias> - the work area (file handle) in which to move the record
pointer.
USE - opens the specified table and its associated index and memo files, if any.
Syntax
USE
[<filename1> [[TYPE] PARADOX | DBASE]
[IN <alias>]
[INDEX <filename2> [, <filename3> ...]]
[ORDER [TAG] <.ndx filename> |
<tag name> [OF <.mdx filename>]]
[AGAIN]
[ALIAS <alias name>]
[EXCLUSIVE | SHARED]
[NOSAVE]
[NOUPDATE]]
<filename 1> - the table you want to open.
IN <alias> - the work area (file handle) in which to open the table. You
can
specify the work area that is being used by another table, in
which
case that other table is closed first.
INDEX <filename2> [, <filename3> ...] - index file
ORDER [TAG] <tag name> -
AGAIN - opens a table and its related index files in the current or
specified
work area, leaving the table open in one or more other work
areas.
ALIAS <alias name> - an alternate alias name to assign to the table.
EXCLUSIVE | SHARED - EXCLUSIVE opens the table so that no other users can
open
the table until you close it; SHARED allows other users access
while
the table is opened.
NOSAVE - used to open a table as a temporary table. When you close a
table
opened with NOSAVE, it is erased along with its associated index
and
memo files. If you inadvertently open a table with the NOSAVE
option,
use COPY to save the data.
NOUPDATE - prevents users from altering, deleting, or recalling any
records in
the table.
INDEX - creates an index for the current table.
APPEND FROM - copies records from an existing table to the end of the current
table.
Syntax
APPEND FROM <filename>
[FOR <condition>]
[[TYPE] SDF
DELIMITED [WITH
<char> | BLANK]]
[REINDEX]
<filename> - the name of the file whose records you want to append to
the current table.
FOR <condition> - restricts APPEND FROM to records in <filename> that
meet
<condition>. You can specify a FOR <condition> only for
fields
that exist in the current table.
SKIP - moves the record pointer in the current or specified work area.
Syntax
SKIP [<expN>] [IN <alias>]
<expN> - the number of records Visual dBASE moves the record pointer
forward
or backward in the table open in the current or specified work
area.
If <expN> evaluates to a negative number, the record pointer
moves
backward. SKIP with no <expN> argument moves the record pointer
forward one record.
IN <alias> - the work area in which to move the record pointer.
GO TOP - move record pointer to first record
GO BOTTOM - move record pointer to last record
GO <recno> - position record pointer to <recno>
RECCOUNT() - returns the number of records in a table.
Syntax
RECCOUNT([<alias>])
<alias> - the work area you want to check.
RECNO() - for DBF tables, returns the current record number.
Syntax
RECNO([<alias>])
<alias> - the work area you want to check.
8. Re: FILEMAN.E
Hi Brian Jackson,
More on common xbase commands.
Arlie Codina
web.master at flashmail.com
*** XBase Common Data Manipulation Commands ***
DELETE - deletes/mark records from the current table.
Syntax
DELETE
[<scope>]
[FOR <condition 1>]
[WHILE <condition 2>]
<scope>
FOR <condition 1>
WHILE <condition 2>
The scope of the command. The default scope is NEXT 1, the current
record only.
DBF tables support the concept of soft deletes, where the record is
marked as deleted and normally hidden when SET DELETED is ON
(the default). If you SET DELETED OFF, you can see the deleted
records along with the records that are not marked as deleted. You
can RECALL the record to undelete it. To actually remove the record
from the table, you must PACK the table. If you use the LIST or
DISPLAY commands to display records, records marked as deleted are
displayed with an asterisk.
PACK - removes all records from a DBF table that have been marked
as deleted.
SET DELETED - controls whether dBASE hides records marked as deleted
in a DBF table.
Syntax
SET DELETED ON | off
Description
Use SET DELETED to include or exclude records marked as deleted
in a DBF table. When SET DELETED is OFF, all records appear in a
table. When SET DELETED is ON, dBASE excludes records that have
been marked as deleted, hiding them from most operations.
INDEX, REINDEX, and RECCOUNT() aren't affected by SET DELETED.
If two tables are related with SET RELATION, SET DELETED ON
suppresses the display of deleted records in the child table.
The related record in the parent table still appears, however,
unless the parent record is also deleted.
RECALL - restores records that were previously marked as deleted
in the current DBF table.
Syntax
RECALL
[<scope>]
[FOR <condition 1>]
[WHILE <condition 2>]
<scope>
FOR <condition 1>
WHILE <condition 2>
The scope of the command. The default scope is NEXT 1, the
current record only.
Description
Use RECALL to undelete records that have been marked as deleted
in the current DBF table with DELETE but have not yet been removed
with PACK. Executing DELETE marks the record as deleted but doesn=92t
physically remove them from the table. If SET DELETED is ON
(the default), these deleted records cannot be seen. RECALL reverses
this process, unmarking the records and fully restoring them to the
table.
Records eliminated with PACK or ZAP are permanently removed and can't
be recovered using RECALL.
When using RECALL, SET DELETED should be OFF; otherwise you will not
be able to see the deleted records you want to recall. Using RECALL
on records that are not marked as deleted has no effect.
ZAP - removes all records from the current table.
Syntax
ZAP
Description
ZAP is the fastest way to delete all records from a table. DELETE
ALL, followed by PACK, also deletes all records from a table.
Using ZAP requires a table be opened exclusively.
