You are not logged in.
Pages: 1
Hi Arnaud,
what do you think about vector-containers? https://community.idera.com/developer-t … for-delphi
Offline
It is a good idea from the C++ point of view.
When the author calls its library an "ORM" it is a joke for sure, since its flat persistence on disk is just awful - especially due to shortstring use. A SQLite3 file is just better.
In practice, I think this library is a niche for his own projects, but not production-ready for everyone. Especially the shortstring for text fields, which is slow, take 256 bytes on memory or disk whatever the length is, is limited to 255 bytes, and not Unicode aware, is a showstopper.
It seems less useful, and less powerful that the records and dynamic arrays wrappers, as available in mORMot.
Our TDynArray wrapper is much more complete (e.g. about sorting/comparing/searching), can have multiple hash tables for O(1) lookups, has much better binary serialization, and features JSON.
It will be even better in mORMot2 - I just rewrite the whole RTTI kernel, and optimized a lot binary and JSON process.
Offline
Reading the article, I couldn't stop thinking about how much better TDynArray is.
And a big Plus: dispenses the use of generics.
Offline
+1 ... :-)
Offline
Thank you for your answer Arnaud and for the classification of the approach with regard to mORMot and its current (and future) possibilities.
By studying mORMot since 2015, its docu, forum, development in code, etc. I learn a lot about good concepts and their modeling; from DDD down to a machine-level implementation.
Yes mORMot seems to be a very good teacher. But also a very very hard one! Especially if you're not a programmer, just a businessman, who has to decide about conceptually correct software development in a small company, it's quite a challenge to see more and more clearly and anticipate the right direction.
So thanks again to Arnaud and everyone else in this great community!
With kind regards
Eduard
Offline
Pages: 1