make default table output a bit more compact

[shua]

I found the default formatting for summarize summarize <file> to be visually noisy and difficult to read because of the separators between each line:

$ summarize summarize my-data.mm_profdata
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| Item                                                            | Self time | % of total time | Time     | Item count | Incremental result hashing time |
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| LLVM_passes                                                     | 714.79ms  | 13.408          | 714.79ms | 1          | 0.00ns                          |
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| run_linker                                                      | 706.56ms  | 13.253          | 706.56ms | 1          | 0.00ns                          |
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| LLVM_module_codegen_emit_obj                                    | 690.58ms  | 12.954          | 690.58ms | 173        | 0.00ns                          |
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| codegen_module                                                  | 401.81ms  | 7.537           | 748.49ms | 172        | 0.00ns                          |

... and so on for like 2 pages of scrolling

This change removes the separating lines +-----+---+ between rows. This introduces a new problem where it is hard to read a line as there's often a lot of whitespace between the end of the item label and the beginning of the next column. I added some ... to pad out the line to make it easier to scan.

$ target/debug/summarize summarize my-data.mm_profdata
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| Item                                                            | Self time | % of total time | Time     | Item count | Incremental result hashing time |
+-----------------------------------------------------------------+-----------+-----------------+----------+------------+---------------------------------+
| LLVM_passes ................................................... | 714.79ms  | 13.408          | 714.79ms | 1          | 0.00ns                          |
| run_linker .................................................... | 706.56ms  | 13.253          | 706.56ms | 1          | 0.00ns                          |
| LLVM_module_codegen_emit_obj .................................. | 690.58ms  | 12.954          | 690.58ms | 173        | 0.00ns                          |
| codegen_module ................................................ | 401.81ms  | 7.537           | 748.49ms | 172        | 0.00ns                          |
| incr_comp_encode_dep_graph .................................... | 302.17ms  | 5.668           | 302.17ms | 223438     | 0.00ns                          |
| implementations_of_trait ...................................... | 280.60ms  | 5.263           | 395.25ms | 44604      | 112.82ms                        |

A similar change has been made for the table output of the summarize diff ... command.

I think the final result is able to present more data in a single screen, while also not sacrificing ease of scanning horizontally.

[bjorn3]

This doesn't seem to have any effect for me locally.
Edit: Seems like you only changed the summarize subcommand, not the diff subcommand. Please also make this change to the diff subcommand.

[TechPizzaDev]

A quick heads up that this is getting very close to the format of tables in Markdown, where it would look like this (replace + with | and remove ... for padding):

| Item                          | Self time | % of total time | Time     | Item count | Incremental result hashing time |
|-------------------------------|-----------|-----------------|----------|------------|---------------------------------|
| LLVM_passes                   | 714.79ms  | 13.408          | 714.79ms | 1          | 0.00ns                          |
| run_linker                    | 706.56ms  | 13.253          | 706.56ms | 1          | 0.00ns                          |
| LLVM_module_codegen_emit_obj  | 690.58ms  | 12.954          | 690.58ms | 173        | 0.00ns                          |
| codegen_module                | 401.81ms  | 7.537           | 748.49ms | 172        | 0.00ns                          |
| incr_comp_encode_dep_graph    | 302.17ms  | 5.668           | 302.17ms | 223438     | 0.00ns                          |
| implementations_of_trait      | 280.60ms  | 5.263           | 395.25ms | 44604      | 112.82ms                        |

Item	Self time	% of total time	Time	Item count	Incremental result hashing time
LLVM_passes	714.79ms	13.408	714.79ms	1	0.00ns
run_linker	706.56ms	13.253	706.56ms	1	0.00ns
LLVM_module_codegen_emit_obj	690.58ms	12.954	690.58ms	173	0.00ns
codegen_module	401.81ms	7.537	748.49ms	172	0.00ns
incr_comp_encode_dep_graph	302.17ms	5.668	302.17ms	223438	0.00ns
implementations_of_trait	280.60ms	5.263	395.25ms	44604	112.82ms

[Kobzol]

I like the compact format, but having it as a Markdown table would be even better.. :)