[whstl]

No, it comes from wanting to make autocompletion easier and to make variable scoping/method ordering make sense within LINQ. It is an actual improvement in this regard.

LINQ popularized it and others followed. It does what it says.

Btw: saying that "people try to impose their own comfort" is uncalled for.

[ltbarcly3]

In that case you are just objectively incorrect, you can build a far, far more efficient autocomplete in the standard query order. I will guess something like half as many keystrokes to type the same select and from clauses. You are imagining a very niave autocomplete that can only guess columns after it knows the tables, but in reality you can guess most of the columns, including the first one, the tables, and the aliases. Names in dbs are incredibly sparse, and duplicate names don't make autocomplete less effective.

If you are right about why they did it its even dumber than my reason, they are changing a language grammar to let them make a much worse solution to the same problem.

[whstl]

An autocomplete that shows only the column names of the desired table BEFORE the from clause is typed by the user would require a time machine.

Sure you can do something that is close enough, but the LINQ authors were looking for precision in the autocompletion and for the LINQ query to have the same ordering as expression syntax.

The goals of this syntax are very precise and people seem to like it. Once again: calling it dumb is uncalled for.

[ltbarcly3]

So you want it to work this way, regardless of how well autocomplete works? Sounds like its about your personal comfort to make it work like another system you are more familiar with, which is exactly what I suggested.

It doesn't require a time machine, just a basic understanding of statistics or probability.

[whstl]

I’m comfortable with pretty much anything. It took me like 2 mins to get used to this syntax in LINQ.

On the other hand, statistical autocomplete is not as good as having a precise autocomplete that does’t require jumping around lines.

My point here is that different people enjoy different things. There is no need to shit on other people’s accomplishments or preferences.

[ltbarcly3]

Name one thing thst uses autocomplete like i am describing.

[pests]

I don’t want to type any column names. When you start with FROM the only autocomplete suggestions available are the columns from the specific table, not the entire database. How many columns do I need to type before you can single down a single table? What if you have multiple tables with the same column names?

[ltbarcly3]

This is extremely easy to check. It depends on the schema.

If your tables have very heterogeneous column names, like 1 column will identify any table on average. There will be some duplicates but the median columns will be one or two, but generally you can even complete those after a few characters.

If your database has very homogenous column names you don't need to identify a single table for autocomplete to be very precise, unless there is no correlation between column name co occurence within tables. However if there is no correlation you are back to very low number of columns to identify the table.