ofc., no code format / refactoring issues until GSOC finish (so I didn't create an issue for that and just started the discussion here).

Now is 21st century, punch cards are gone and almost any popular data-displaying device could easily display 110 and even 120 symbols per line.
For example, my Android device (5'') seems to display 120 symbols-per-line code well in album mode. In portrait mode it is hard to read code even on my Nexus 7'' (even if code formatted to 80 symbols per line - it is similar to programming on PC with portrait-mode display)
But.. After I tried to print 120-symbols code directly from Eclipse, I noticed that 110 - symbols lines gets wrapped when printing code from Eclipse using default font and margins. There is no way to reduce margins, the only thing is to reduce editor's font size, print and then revert font size back (but this is an ugly trick).
So OK, lets use 100 symbols as max like length. In any case, 100 symbols is much better that 19th century format of 80 symbols ;)
2. Please, look at Sergey message and take a look at his examples first, then I will provide mine, if they will be still needed.
3. Lets me share some thoughts about this:
Checkstyle Checks could be easily configured in a bad manner which will lead to ugly code ..
But I don't think that it is the reason to configure Checkstyle in a bad (or very-very and very unpopular) manner and just use it for development to 'eat own food' ;)
Let me explain you the reason why I am OK to remove that prefixes:
c. I did not find any actual opensource Java project which uses prefixes notation (ofc., except for Checkstyle with it's unique code convention).
Code printed directly from IDE keeps coloring and so there is no need to keep prefixes on fields as they will be marked with blue color.
For me that prefixes does not help to review code on GitHub or Android as I don't remember what prefix stands for fields and what for method parameters ) I always forget about 'm' stands for 'member' and that there is no 'f' prefix for fields, and also about method parameter prefix is 'a' instead of 'm' which would be more convenient for me =)
Convenience is very subjective term, so I would agree to ignore Oracle and Google code conventions about prefixes for convenience of somebody in team, but only if anybody else from project admins / contributors think that these prefixes are convenient.
Lets vote.
4. I am not sure that we need 'package-info' files for javaDoc generation at sevntu-checkstyle currently. Lets investigate farther.
=======================================================================
Let me also remind that currently sevntu-checkstyle project does not match our own 'Checkstyle For Checkstyle' config at all ) Brief example:
ForbidThrowAnonymousExceptionsCheck.java - no prefixes at all variable names.
So, maybe, lets include checking of sevntu-checkstyle project with 'Checkstyle For Checkstyle' config into maven build (in order to enforce the same code convention on all Checkstyle sources)? Only after that we will really start to 'eat our own food'.
But before fixing 'Checkstyle For Checkstyle' config warnings at sevntu-checkstyle it would be ofc. good to clear that config from some severe restrictions if most part of team does not consider them right.
I think, it should be an iterative process like this:
1. Enforce automatic checking of sevntu-checkstyle code with 'Checkstyle For Checkstyle' config.
2. Discuss some warnings on real-world examples and reduce restrictions, where needed or fix warnings.
3 Do '3' until all Checkstyle warnings gone.
4. Create the PR containing 1-3 code updates.