Improve quality of string output about gofuzz

[lavalamp]

I think making the chance of getting an empty string configurable is desirable.

The character distribution is tricky; ideally we'd want to test characters of all UTF-8 encoding forms; that's why it is how it is. I agree spaces would be desirable.

As far as duplicates-- right now there's just a small chance (5% IIRC) of it outputting a blank string. That's on purpose, since presumably there's a lot of different places to store these strings, and we eventually want to try a blank in all of them.

[Fresheyeball]

I agree the distribution is tricky, all UTF-8 is desirable. I just think it can be done much better, perhaps interleaving different sources for randomness. Using a wikipedia article as a source, as well as other texts in addition to pure UTF8 could be useful. I've outputted strings from QuickCheck and generally had more desirable results. Perhaps something in there can lead the way.

As for blank strings and dealing with duplicates. I think ideally each instance of fuzz should maintain internal state to prevent duplicates. This way the odds of encountering an empty sting can be 100% or 0% while ensuring that it does not appear more than once (running the same input more than once is a wasted test).

Just brainstorming, I think it would be desirable to configure the presence of empty stings (I generally like them), but also the guaranteed presence of special strings like \n. I'd sleep better knowing that those are always present in the input of tests.

More variance on string length as well would be appreciated.

[Fresheyeball]

I ran a sample from QuickCheck if you are interested:

����������(50 tests)����������}�{O"(�
����������(51 tests)����������OLÍ�F�fÅpù�KÏG'e}j
v^ ésÈ.Rfp(Ì��ð��P�L@�dRj
����������(52 tests)����������L�bdXC�(rüLgY�;9A�aPáb%o×�
����������(53 tests)������������a
����������(54 tests)����������4åÕ�[7jT��ç|¶od
[y  �"�~2�ý�"1w��ÄA
����������(55 tests)����������PE¢�Øf[�_���
����������(56 tests)����������a��<s\#6��4\.·/H.Pbw�u0
����������(57 tests)����������ñºKi�F94¿�\��1$­
����������(58 tests)�����������93�\ Ææ�=R×�ò!g�Öt��Ï
����������(59 tests)����������]-'5�Q@kúH�M�%óRz=MM/�¹¾ì�qv��<Þ��$:ñ-�îR�7NUL=/
����������(60 tests)����������>%f~A~�%8IE�å�~!2#�LJiðhî÷�¢B¯Æ�q��RZfb��=3c~4óuÄ>¹3�
����������(61 tests)����������O�GÀX�
����������(62 tests)����������7_��wu pÙNçÁ;@R
����������(63 tests)����������So¿D1;YÕC8�N�ggu�Llpz!�é=�g%þ�
����������(64 tests)����������+
����������(65 tests)����������­f+ôO�%¼�_7Æ×w!
����������(66 tests)����������Â�¶a
l�à
����������(67 tests)����������LAÞxbj�L�*^GcG#_oV�~¯I�"�1y¢��DÆ3
�Z2£2b��`Wp7JX{�µ
����������(68 tests)����������^Ñ95���ü
C��ûß���ñ+�½¡¬)�::qüó�Ü'$ 3IÎ]0nRÎDO
����������(69 tests)����������i'�LAÓ¡s§kÔ)p�þ�zIÁwvT�X
����������(70 tests)�����������
����������(71 tests)����������~��s% ã�(�m7&,_f�g÷
0f7"MNX3JFýÇS/���':'�%��òåG[�¾��
����������(72 tests)�����������mà�£Ø#3�Ú6á :Ùk�Zzj\$��B
P�·T,�F
����������(73 tests)����������R��¥�?ww�Z�ã
����������(74 tests)����������4âWÊd¯)îEB�ß�Z�6Ú�]1ß��<Oy�¨Ã�1.�-�åp�Á\¾·Ub3�g��¨
����������(75 tests)�����������iS8
�'<£u^
�IKA�&¹Q(N³�5�A×û÷A4Wj/Fz6
����������(76 tests)�����������S��â"AtÀ�s6���Zj���Ò!1�ùæû�&BK=#b«Úû[DìØ!
����������(77 tests)����������  C¯Y·��üè�;�ä�
����������(78 tests)����������Äa>{P��h.3$¤�ñbrgr.y�
gk9�9�\H�ÑQ�%r�(&,ò
G�ngä7fó
����������(79 tests)������������&/r33^%Bm)V�%z[N� ¢9ÝH��/·mL"
k�Q�Ca_kÛ�L�FK1#G�ö�K06 c¬�{[
����������(80 tests)����������7ö�Q�#ry1L

[Fresheyeball]

Great lib by the way. I'm doing all my testing with gofuzz