walles/moar
1
1
Fork 0
mirror of https://github.com/walles/moar.git synced 2025-01-07 13:56:33 +03:00
Code Issues Projects Releases Wiki Activity
371 Commits 111 Branches 127 Tags 6.9 MiB
de4028e7b7
Commit Graph

371 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Johan Walles
de4028e7b7
Merge pull request #58 from 89z/master 
Simplify StyleDefault definition
 2021-05-11 15:22:40 +02:00
Steven Penny
f76fd6ddd3 Simplify StyleDefault definition 
StyleDefault is all zero values, so we can simplify the definition. If that
changes in the future (not likely) we can go back to the old definition method.

Fixes #57
 2021-05-11 08:14:15 -05:00
Johan Walles
5e49329110 Split a large function into smaller ones 
For great readability.
 2021-05-08 17:21:08 +02:00
Johan Walles
e3e8be0b8b CTRL-L isn't a thing any more 
Since replacing tcell with twin, we now redraw the whole screen every
time, and CTRL-L doesn't do anything.
 2021-05-08 15:05:12 +02:00
Johan Walles
5d5ecae7c0 Speed up line counting 4x 
And improve large file loading performance in the process.
 2021-05-08 14:16:38 +02:00
Johan Walles
09ab8a4072 Add a line counting performance benchmark 2021-05-08 13:57:54 +02:00
Johan Walles
78ab63273a Move some test-only code into *_test.go files 2021-05-08 10:32:04 +02:00
Johan Walles
99d8db63fb Improve startup interactivity 
By moving line counting into the background goroutine that consumes the
input stream.

This will make the UI appear faster for large files where line counting
takes time.
 2021-05-08 10:26:16 +02:00
Johan Walles
5e847f4cd7 Improve large file reading performance by 10% 
By reading the file in two passes.

The first pass just counts the lines in the file.

The second pass starts by preallocating the correct number of lines, and
then reads the file.

The whole time saving comes from not having to dynamically resize the
lines slice while reading the file.
 2021-05-03 20:33:25 +02:00
Johan Walles
97884d1805 Add a large file benchmark 2021-05-03 19:15:43 +02:00
Johan Walles
f8e45f73e2 Log input stream consumption time 
At debug level.
 2021-05-03 08:08:41 +02:00
Johan Walles
9af12f451c Move code outside of a lock 2021-05-03 06:44:57 +02:00
Johan Walles
7a66c633ca Lower memory usage by 10% 
When loading a 577MB file.
 2021-05-02 19:28:51 +02:00
Johan Walles
306442f2b2 Split a line to improve profilability 
The line is top of the pprof alloc profile, but splitting it might
show which allocation on that line it is that is so common.
 2021-05-02 17:36:22 +02:00
Johan Walles
96d26bb97c Avoid some memory allocations 
By reusing the same slice for reading all lines from the input file.
 2021-05-02 14:47:20 +02:00
Johan Walles
3f4113e7bf Log trace events on user input 2021-04-29 23:04:12 +02:00
Johan Walles
7c553891f7 Fix a progress message 2021-04-27 07:05:52 +02:00
Johan Walles
e63871ae6a Make BenchmarkPlainTextSearch() 15+% faster 
By shortcutting styledStringsFromString() on strings without ESC
characters.
 2021-04-26 20:06:08 +02:00
Johan Walles
ef612a5620 Make BenchmarkPlainTextSearch() 50+% faster 
By using a strings.Builder rather than roll-our-own to build a plain
string.
 2021-04-26 08:55:42 +02:00
Johan Walles
860dcd22b4 Make BenchmarkPlainTextSearch() 30% faster 
By taking a cheap shortcut on strings without backspace characters.
 2021-04-26 06:53:30 +02:00
Johan Walles
f5273fc4b3 Make BenchmarkPlainTextSearch() 20% faster 
Using shortcuts on lines without backspace characters.
 2021-04-26 06:40:30 +02:00
Johan Walles
7520c9fc8a Make BenchmarkPlainTextSearch() 12% faster 
By optimizing plainification for all-ASCII strings.
 2021-04-26 00:24:21 +02:00
Johan Walles
02d66fb31c Merge branch 'walles/quicksearch' 
This makes BenchmarkPlainTextSearch() report 30% quicker iterations.
 2021-04-25 23:53:25 +02:00
Johan Walles
a5f43e3d2f Do plaintext like we do cells 
This loses a lot of the performance improvement originally intended, but
this is still 30% faster than before.

Good step in the right direction.
 2021-04-25 23:50:14 +02:00
Johan Walles
d6d428f0ef Plaintext broken UTF8 as ? 
Just like we render it, but without the coloring.
 2021-04-25 23:38:59 +02:00
Johan Walles
17d8a0372a Render unprintable characters as highlighted '?' 2021-04-25 23:35:47 +02:00
Johan Walles
7bf168255a Make test read files like the pager 2021-04-25 21:58:31 +02:00
Johan Walles
e7995bc40f Use ? to mark unprintable chars in plain text rendering 2021-04-25 20:40:38 +02:00
Johan Walles
1d93f43838 Test: Accept prettified man page bullets 2021-04-25 16:32:00 +02:00
Johan Walles
b17c44af28 Fix off-by-X problems comparing strings 
Due to byte offsets vs rune offsets.

This fixes some invalid test failures.
 2021-04-25 13:21:03 +02:00
Johan Walles
5e59515d3f Improve new test diagnostics 2021-04-25 12:25:52 +02:00
Johan Walles
7814f6af2f Add more plaintext-vs-cells verification 2021-04-25 10:08:19 +02:00
Johan Walles
5e2a63dc36 Handle backspace while plaintexting 
More tests pass now, still not all though.
 2021-04-24 23:15:53 +02:00
Johan Walles
66175f02aa Initial tab expansion 
Tests still fail, but not as many.
 2021-04-24 20:49:58 +02:00
Johan Walles
2c20fc31fe Special case stripping string formatting 
Stripping string formatting is on the hot path while searching. This
change makes BenchmarkPlainTextSearch() over 7x faster.

But it also has problems with tab expansion so some tests fail, let's
see how we should handle that.
 2021-04-24 17:20:09 +02:00
Johan Walles
c8a8cb4517 Improve memory usage for large files 
Test:
* Open a 35MB text file
* Search it for some string it doesn't contain

Memory usage before this change: 2.3GB
Memory usage after this change:  0.5GB

The trick is to only provide cells from Lines on demand, since we only
need a few of them when scrolling.

And not storing all of those cells is where the gain in memory usage
comes from.
 2021-04-24 16:27:54 +02:00
Johan Walles
c8bc7b0161 Don't highlight files larger than 1MB 
They make the highlighter just run in the background slowly eating up
memory.
 2021-04-24 16:14:30 +02:00
Johan Walles
9e6af21106
Merge pull request #52 from 89z/master 
DeInit: make output copy friendly
 2021-04-24 09:22:14 +02:00
Steven Penny
63082a804d m/embed-api.go: fix formatting 2021-04-23 19:33:25 -05:00
Steven Penny
374378b979 DeInit: make output copy friendly 
With DeInit turned off, the current output cannot be usefully copied, as no
newlines are emitted. Close the screen in all cases now, but if DeInit is
disabled, reprint the current buffer with newlines after the screen is closed.

Fixes #51
 2021-04-23 19:24:05 -05:00
Johan Walles
d8644ed8e1 Add search performance benchmarks 2021-04-23 22:10:59 +02:00
Johan Walles
60b59c2483 Only install linters when needed 
This makes most runs of ./test.sh quicker since they don't need to
invoke "go install".
 2021-04-23 18:38:40 +02:00
Johan Walles
86ae632f14 Merge branch 'walles/async-highlight' 
Do highlighting in the background. For large files, or files that are
slow to highlight, this removes the initial highlighting delay.
 2021-04-23 06:52:56 +02:00
Johan Walles
232b784428 Fix tests 2021-04-22 22:05:36 +02:00
Johan Walles
c7a26137ef
Merge pull request #50 from 89z/master 
Use CONIN$ for Windows screen setup
 2021-04-22 21:39:14 +02:00
Steven Penny
0f9ae758d5 Use CONIN$ for Windows screen setup 
Fixes #49
 2021-04-22 14:03:22 -05:00
Johan Walles
2c3cc80500 Don't use the plaintext highlighter 2021-04-22 19:31:47 +02:00
Johan Walles
095693eb0b Wait for highlighting to complete in pager tests 2021-04-22 19:26:29 +02:00
Johan Walles
488aaaab88 Highlight in the background 
With this change in place, we first quickly get the file contents
without highlighting. Then, when the highlighting is done, the view gets
updated with the highlighted version.

Useful when you're loading long files where highlighting can take time.
 2021-04-22 19:26:29 +02:00
Johan Walles
1359a3e558 Merge branch 'walles/test-mod-tidy-install' 
Fixes #48 by doing "go mod tidy", getting linters using "go mod install"
and bumping to Go 1.16 to get everything working in CI.

Thanks for your research @89z!
 2021-04-22 19:23:15 +02:00