Hello,

In the following code (
https://github.com/ik5/redit_client.fpc/blob/master/tests/test_parser.lpr#L55),

when I point, I have access violation when trying to create TRedisErrorReturnType.Create

It does not happen for TRedisStatusReturnType.Create for example, that have the same parent as TRedisErrorReturnType.

What am I missing or doing wrong here ?

Thanks,

Ido

Solved it, but I do not understand it.
Why did new made access violation and StrAlloc not ?

Because new() just allocated one pchar and not a buffer. Why do you think using a pchar like this is faster than using string? Have you benchmarked it?

Henry

I haven't benchmark anything in my life :)
PChar is an array like approach. AnsiString, is a record based pointer. It uses more memory, and provide a bit overhead for the same thing. At least on theory. I would love to learn that it's not the case.

Thanks

Ido

ik wrote

PChar is an array like approach. AnsiString, is a record based pointer.

AnsiString also stores the string in an array. It is dynamic, though, so it stores the size and also reference count.

It uses more memory, and provide a bit overhead for the same thing. At least on theory. I would love to learn that it's not the case.

The only thing you save on here is reference counting, which is probably negligible. That's why I asked if you benchmarked it. Especially since you've already created a bug in trying to use pchar.

Henry

PChar is an array like approach. AnsiString, is a record based

pointer. It uses more memory, and provide a bit overhead for the same thing.

At least on theory. I would love to learn that it's not the case.

AnsiString stores and keeps track of the string length and this has many advantages:

*) If you need the length of a Pchar string you always need to count character by character (until #0). This is very slow. Ansistring has it available directly.

*) If you compare two strings for identity you can first compare lengths. If different then strings are different too. Not possible with pchar.

*) You can store #0 in AnsiStrings (not possible with pchar because it indicates the end of the string).

*) You can access the last string character directly by using the length. With pchar you have to iterate on all characters.

Also you don't need to manually manage the storage, it will shrink and grow automatically as you append / delete it.