On 12/10/19 11:08 AM, Laszlo Kishalmi wrote:
Well I'd:
1. convert tab to spaces (however I hope no one using tabs by now)
2. Create a histogram from the leading spaces
3. Start some division test.
Let say we test with indent: IND
for each all leading space number in the histogram with IND then
multiple the remainder with the number of lines which that leading
space is used that would be DEV
summarize of these DEV values (deviations)
The deviation for IND=1 shall be 0. Start to increase IND, the largest
number while IND is 0 the indent of the file.
Actually do not just increase the IND but test a few like 2-8 spaces and
where the max IND which below that threshold would be the detected
indenting.
For miss formatted files you might allow a threshold somewhat above 0
as well. This threshold could be a function of the number of lines in
the file.
On 12/10/19 10:50 AM, Emilian Bold wrote:
Well, this is some original idea! If somebody needs help with such a
thing, I would do it just to learn.
But honestly for an existing file I believe you could detect the
indentation style just looking at the AST. Heuristics would work imho.
But, of course, a neural network would be the cool way of solving it.
--emi
On Tue, Dec 10, 2019 at 8:44 PM Siddhesh Rane
<[email protected]> wrote:
Glad to see someone interested in this feature.
A long time ago I came across a blog post[1] by Andrej Karpathy (AI
guru) in which a neural network was trained on variety of text
datasets and then used as a content generator. When trained on Linux
and Apache source code, the generator could reproduce pseudo code,
that was not valid code, but was very well formatted. It was quite
cool to see the network had learned several things, indentation
being one of them. Do check out the Source Code section of the blog
to know what I am talking about.
The most straightforward way I see to do this is some frequency
model of a predefined set of indentation rules.
But if anyone is interested to create a small fast neural network
with limited features, it would be great to see the possibilities.
+1 for this.
Regards
Siddhesh
[1] http://karpathy.github.io/2015/05/21/rnn-effectiveness/
December 10, 2019 11:16 PM, "Alvin Thompson"
<[email protected]> wrote:
Hi,
Assuming that this email doesn't get flagged as spam due to the
subject line, this is just a
reminder that I have a bounty for anyone who creates a NetBeans
plugin that detects and uses the
existing indentation of the file being edited. For any indentation
info that can not be gleaned
from the file, it should fall back to the preferences. This really
is an absolute must team work
these days.
You get $100 for coming up with the plugin, and another $100 if it
gets donated and makes it into
NetBeans itself. But I imagine most people will want to do this so
that I have a slightly higher
opinion of them.
I would do this myself but I am very, very lazy.
That is all,
Alvin
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists