Re: Formatting a number and digits of precision - C# Discussion Boards

6-Nov-18 12:19

I need to have 15 digits of precision (after the decimal point) on a double precision value when it's written to a file for comparison to some legacy code. I can't seen to get it work consistently.

For instance if I perform the following;

double d = 0.123456789012345
Debug.WriteLine(d.ToString("G15"));

I get the expected output of -> 0.123456789012345 15 trailing digits of precision as expected

Then if I have the following;

double d = 10.123456789012345;
Debug.WriteLine(d.ToString("G15"));

The output is -> 10.1234567890123 13 digits of trailing precision

Finally;

double d = 12345.123456789012345;
Debug.WriteLine(d.ToString("G15"));

The output is -> 12345.123456789 9 digits of precision after decimal point

Using good old C++ you can just specify "%.15lf" as a format specifier and always get 15 digits of precision. This just seems to give a specific number of digits based shifting the decimal point.

Mycroft Holmes6-Nov-18 13:28

Mycroft Holmes

6-Nov-18 13:28

It seem to me that "G15" produces 15 digits, not 15 decimal places!

Never underestimate the power of human stupidity -
RAH
I'm old. I know stuff - JSOP

OriginalGriff6-Nov-18 21:01

6-Nov-18 21:01

Try:

d.ToString("F15")

But...Double values only have a precision of "15 ~ 17" digits (depending on the value, it's down to how they are stored as binary): double keyword (C# Reference) | Microsoft Docs[^] so specifying 15 digits after the decimal poitn is probably not going to work in many cases.

Sent from my Amstrad PC 1640
Never throw anything away, Griff
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!

csharpsucks7-Nov-18 4:32

7-Nov-18 4:32

Using "F15" produces the same result, this is maddening.

Richard Deeming7-Nov-18 4:38

Richard Deeming

7-Nov-18 4:38

OriginalGriff wrote:
But...Double values only have a precision of "15 ~ 17" digits (depending on the value, it's down to how they are stored as binary)... so specifying 15 digits after the decimal point is probably not going to work in many cases.

I suggest you read that part again. Smile | :)

If the total precision of the number is 15-17 digits, and you have four digits before the decimal point, you're not going to be able to get 15 digits after the decimal point.

"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer

OriginalGriff7-Nov-18 4:44

7-Nov-18 4:44

I get exactly what I expect:

double d = 0.123456789012345;
Debug.WriteLine(d.ToString("G15"));
d = 12345.123456789012345;
Debug.WriteLine(d.ToString("F15"));

0.123456789012345
12345.123456789000000

Remember what I said?

Quote:
But...Double values only have a precision of "15 ~ 17" digits (depending on the value, it's down to how they are stored as binary): double keyword (C# Reference) | Microsoft Docs[^] so specifying 15 digits after the decimal poitn is probably not going to work in many cases.

If the double variable can only hold a total of 15~17 significant decimal digits, then you can't get 20 output of it!
Remember how double values are stored: as binary with a mantissa and an exponent. So the code (rightly) prints the same number of significant digits and pads the rest to the size you asked for.

To be honest, if you think any number you can come up with is going to be accurate to fifteen digits, you are very probably wrong! Laugh | :laugh:

Sent from my Amstrad PC 1640
Never throw anything away, Griff
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!

csharpsucks7-Nov-18 6:08

7-Nov-18 6:08

Oh i totally agree with you on the accuracy thing.

Oddly enough in C when you do the same thing;

double d = 0.123456789012345;
char buf[20];
sprintf(buf, "%.15lf\n", d);
OutputDebugStringA(buf);

output is 0.123456789012345

d = 12345.123456789012345;
sprintf(buf, "%.15lf\n", d);
OutputDebugStringA(buf);

output is 12345.12345678901

I still get 2 more digits of precision and that's exactly what I'm fighting against.

OriginalGriff7-Nov-18 6:28

7-Nov-18 6:28

Not really - The C version depends on the compiler. If I try with an online C compiler[^] I get:

0.123456789012345                                                                                                                                            
12345.123456789011470

If I try with Virtual-C[^] I get this:

0.123456789012345
12345.123456789011135

IT's not "extra accurate digits" it's "different interpretation of the output". Because floating point numbers don't translate well into binary (for example, 0.2 in decimal is a fixed value, in binary is a repeating sequence so it isn't accurate) you can "generate" more digits if you take the first one, then multiple by 10.0 and repeat. That doesn't mean that the values to get are "more accurate" - in fact they are prone to reduced accuracy but a higher digit count. Try it again with some different numbers and you should see what I mean.

Generally speaking, all systems use the same floating point storage design: IEEE 754, (because if they don't they can't use floating point coprocessors to do the hard stuff!)
IEEE 754 - Wikipedia[^]

Sent from my Amstrad PC 1640
Never throw anything away, Griff
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!

csharpsucks7-Nov-18 6:44

Problem at inserting a DB entry

7-Nov-18 6:44

I've punted on this issue it will be interesting to see the results of the data analysis when I load this data back into the system and process against the original data. I'm dealing with nanometer measurement data.

Thanks for the help.

User 136751145-Nov-18 23:49

User 13675114

5-Nov-18 23:49

Hi Guys,

Below code don't allow to create entry at database with null areas and pop-up comes to warn user.

I fill all fields/create entry and all fields are cleared up for new entry but once I try to add new entry, program allows me to create with empty fields.

Do I need to add some code in order form to reload or something like that?

What would be the way of getting rid of this issue?

Thanks.

note: something other than Application.Restart(). that would be great if no close&open app.

private void btnStokEkle_Click_1(object sender, EventArgs e)
        {
            if (txtAdi.Text == "" || txtModel.Text == "" || txtSeriNo.Text == "" || gon.Text == "" || tah.Text == "")
            {
                MessageBox.Show("Lütfen Tüm Alanları Doldurunuz.", "Hata", MessageBoxButtons.OK, MessageBoxIcon.Error);
                return;
            }

            else

            vt.sqlCalistir("insert into kolon (siparis,malzeme,takip,gonder,tahmini,gercek) values ('" + txtAdi.Text + "','" + txtModel.Text + "','" + txtSeriNo.Text + "','" + gon.Text + "','" + tah.Text + "', '" + tt.Text + "')");
            MessageBox.Show("İşlem Kaydı Yapıldı.", "İşlem Tamam", MessageBoxButtons.OK, MessageBoxIcon.Information);

            getir();
            txtId.Clear();
            txtAdi.Clear();
            txtModel.Clear();
            txtSeriNo.Clear();
            gon.EditValue = " ";
            tah.EditValue = " ";
            tt.EditValue = " ";
            txtSeriNo.Text = Environment.UserName;

        }

modified 13-Feb-19 21:02pm.

OriginalGriff6-Nov-18 0:08

6-Nov-18 0:08

Not like that! Never concatenate strings to build a SQL command. It leaves you wide open to accidental or deliberate SQL Injection attack which can destroy your entire database. Always use Parameterized queries instead.

When you concatenate strings, you cause problems because SQL receives commands like:

SQL

SELECT * FROM MyTable WHERE StreetAddress = 'Baker's Wood'

The quote the user added terminates the string as far as SQL is concerned and you get problems. But it could be worse. If I come along and type this instead: "x';DROP TABLE MyTable;--" Then SQL receives a very different command:

SQL

SELECT * FROM MyTable WHERE StreetAddress = 'x';DROP TABLE MyTable;--'

Which SQL sees as three separate commands:

SQL

SELECT * FROM MyTable WHERE StreetAddress = 'x';

A perfectly valid SELECT

SQL

DROP TABLE MyTable;

A perfectly valid "delete the table" command

SQL

--'

And everything else is a comment.
So it does: selects any matching rows, deletes the table from the DB, and ignores anything else.

So ALWAYS use parameterized queries! Or be prepared to restore your DB from backup frequently. You do take backups regularly, don't you?

Sent from my Amstrad PC 1640
Never throw anything away, Griff
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!

User 136751146-Nov-18 3:12

User 13675114

6-Nov-18 3:12

Well noted OriginalGriff. As a C# beginner, I was trying to practice basic CRUD in Access db and yes backup is on.

I am just stuck at solving this. so is it all about SQL commands? if I use correct commands then that click event won't save any entry with a null field?

modified 13-Feb-19 21:02pm.

Luc Pattyn6-Nov-18 3:51

Luc Pattyn

6-Nov-18 3:51

What null field? A TextBox never contains null, if it is empty, then its .Text property returns the empty string, not null.

Smile | :)

Luc Pattyn [My Articles] Nil Volentibus Arduum

tranthanhtu.vn6-Nov-18 18:12

6-Nov-18 18:12

Hi Al Soyo,
This was interesting, Please ping me on skype (tranthanhtu83) if you still not solve this issue.

Best regards,
TU Tran
Technical Leader
Blog: http://tranthanhtu.vn
Mail: contact@tranthanhtu.vn
Mobile: +84 90 883 884 6
Skype: tranthanhtu83
CodeProject: @techcoaching
LinkedIn: tutrancoaching
Github: techcoaching

Richard Deeming7-Nov-18 0:59

Richard Deeming

7-Nov-18 0:59

You've been here long enough to know that this is NOT how CodeProject works.

If you want to help someone, then help them here on the site, so that others who are having the same problem can see the discussion and benefit from it.

"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer

tranthanhtu.vn7-Nov-18 1:27

7-Nov-18 1:27

Hi Richard Deeming,
I intend to check the code and will write another comment for the root cause.
Just look at the code, seems not create such problem.

And I will appreciate if you can talk in more friendly way (such as: this will violate the rule, .....).
We are here to learn and share friend and you are not my boss.

Thanks

Richard Deeming7-Nov-18 1:32

Richard Deeming

7-Nov-18 1:32

Trying to take the discussion off-line violates the rules of this site.

If you want to help people on CodeProject, then help them HERE, not in a Skype discussion.

You've been here over 11 years. That's more than long enough to know the rules!

"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer

tranthanhtu.vn7-Nov-18 1:36

7-Nov-18 1:36

Thank Richard Deeming,

Eddy Vluggen7-Nov-18 1:44

Eddy Vluggen

7-Nov-18 1:44

Since I'm not your boss, I need not stay friendly. There is nothing "interesting" about his "issue", and it seems like spamming. Going for a coffee before I decide whether or not to report this as spam, since your post does not contribute anything.

Bastard Programmer from Hell Suspicious | :suss:

If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.

tranthanhtu.vn7-Nov-18 2:29

7-Nov-18 2:29

Thank Eddy Vluggen for letting me know your idea.
Yeah, it was up to you.
Let consider the situation, @Eddy Vluggen has a problem with his code and the code seems works fine.
The problem was still there, I intend to ask him and may run the code to see any potential cause raises this problem, and may help him to fix. of course, I can write another comment about the root cause if I can found.
And I was not allowed and may be reported as spammer just because someone did not see what I see: The code looks ok but not work.
@Eddy Vluggen: Do you think this is the interesting point.

Again, thank for let me know your idea and you can do what ever you think is right.

Eddy Vluggen7-Nov-18 2:37

Eddy Vluggen

7-Nov-18 2:37

tranthanhtu.vn wrote:
And I was not allowed and may be reported as spammer just because someone did not see what I see: The code looks ok but not work.

The code does not look "ok".

It is also not a particular complicated subject, there's literally examples on MSDN that you can copy and paste.

Bastard Programmer from Hell Suspicious | :suss:

If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.

tranthanhtu.vn7-Nov-18 2:42