url: add urlSearchParams.sort() #11098

TimothyGu · 2017-02-01T08:11:21Z

The URL Standard requires sort() to be stable, which precludes us from using the V8-native sort() function.

I originally wanted to make the function a simple insertion sort only, but I realized it had the potential for a DoS attack because of its nature of being O(n²).

The algorithm for the function follows the norm: insertion sort for small arrays, divide-and-conquer (in this case, a stable merge sort) for larger arrays. In this specific case, the cutoff is set at 59 items, though if performance improvements are made to either sort the cutoff may be adjusted.

Fixes: #10760
Ref: whatwg/url#26
Ref: whatwg/url#199
Ref: web-platform-tests/wpt#4531

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
tests and/or benchmarks are included
documentation is changed or added
commit message follows commit guidelines

Affected core subsystem(s)

url

jasnell · 2017-02-01T16:48:52Z

@mscdex ... I would appreciate you taking a look at this from a perf point of view

jasnell

Almost there

jasnell · 2017-02-01T16:50:19Z

doc/api/url.md

+with a [stable sorting algorithm][], so relative order between name-value pairs
+with the same name is preserved.
+
+This method can be used, in particular, to increase cache hits.


An example may be helpful here.

jasnell · 2017-02-01T16:51:37Z

lib/internal/url.js

+    }
+
+    // arbitrary number found through testing
+    if (len < 118) {


We should look to benchmark this on a number of different systems to see if this number holds up.

jasnell · 2017-02-01T16:57:45Z

test/parallel/test-whatwg-url-searchparams-sort.js

+const {URL, URLSearchParams} = require('url');
+
+[
+  {


Perhaps also include a test with an empty key? e.g. z=a&=b&c=d

joyeecheung

The sorting bit looks like C++ code itself really, and it creates two holey arrays, so probably worth just implement those in C++?

joyeecheung · 2017-02-01T18:37:45Z

lib/internal/url.js

+
+    update(this[context], this);
+
+    function merge(out, start, mid, end, lBuffer, rBuffer) {


This can be moved outside of sort

TimothyGu · 2017-02-01T21:47:37Z

@joyeecheung said:

so probably worth just implement those in C++?

Just tried doing something like that:

C++ diff

diff --git a/src/node_url.cc b/src/node_url.cc
index 0d5e695..3ab7108 100644
--- a/src/node_url.cc
+++ b/src/node_url.cc
@@ -1389,6 +1389,49 @@ namespace url {
                             v8::NewStringType::kNormal).ToLocalChecked());
   }

+  static void SortParams(const FunctionCallbackInfo<Value>& args) {
+    Environment* env = Environment::GetCurrent(args);
+    CHECK_GE(args.Length(), 1);
+    CHECK(args[0]->IsArray());
+
+    Local<Array> src = args[0].As<Array>();
+    uint32_t len = src->Length();
+
+    std::vector<std::string> a;
+    Copy(env, src, &a);
+    std::vector<std::string> tmp(len);
+
+    for (uint32_t step = 2; step < len; step *= 2) {
+      for (uint32_t start = 0; start < len - 2; start += 2 * step) {
+        uint32_t mid = start + step;
+        uint32_t end = mid + step;
+        end = end < len ? end : len;
+        if (mid > end)
+          continue;
+
+        uint32_t l = start;
+        uint32_t r = mid;
+        uint32_t t = start;
+        while (l < mid && r < end) {
+          if (a[l] <= a[r]) {
+            tmp[t++] = a[l++];
+            tmp[t++] = a[l++];
+          } else {
+            tmp[t++] = a[r++];
+            tmp[t++] = a[r++];
+          }
+        }
+        while (l < mid)
+          tmp[t++] = a[l++];
+        while (r < end)
+          tmp[t++] = a[r++];
+      }
+      tmp.swap(a);
+    }
+
+    args.GetReturnValue().Set(Copy(env, a));
+  }
+
   static void Init(Local<Object> target,
                    Local<Value> unused,
                    Local<Context> context,
@@ -1398,6 +1441,7 @@ namespace url {
     env->SetMethod(target, "encodeAuth", EncodeAuthSet);
     env->SetMethod(target, "domainToASCII", DomainToASCII);
     env->SetMethod(target, "domainToUnicode", DomainToUnicode);
+    env->SetMethod(target, "sortParams", SortParams);

 #define XX(name, _) NODE_DEFINE_CONSTANT(target, name);
     FLAGS(XX)

It's much slower than the JS version: in fact, the two Copys alone are several times slower than the JS sort(). I suspect I might be doing something gravely wrong, but I'm not very interested in pushing this further. Feel free to try to improve on this though.

TimothyGu · 2017-02-01T23:14:27Z

@jasnell said:

+    // arbitrary number found through testing
+    if (len < 118) {
We should look to benchmark this on a number of different systems to see if this number holds up.

Agreed. I've made a simple benchmark specifically for this purpose.

I'm unfortunately not too well-versed in data analysis languages, so my method of measuring is rather crude. A regression model is probably better-fit for this purpose.

How to run

# create `merge` and `insertion` binaries, with insertion and merge sort blocks resp.
# commented out

# wait. it's gonna run for a while (~16 minutes)
$ node benchmark/compare.js --old ./insertion --new ./merge --filter rrrr --runs 10 url > sort.csv
$ Rscript benchmark/compare.R < sort

My results

                                                      improvement confidence      p.value
 url/url-searchparams-sort-rrrr.js n=100000 type="02"    -27.63 %        *** 2.033434e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="03"    -36.96 %        *** 5.757754e-07
 url/url-searchparams-sort-rrrr.js n=100000 type="04"    -43.56 %        *** 1.783939e-07
 url/url-searchparams-sort-rrrr.js n=100000 type="05"    -45.19 %        *** 7.799419e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="06"    -50.19 %        *** 1.354454e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="07"    -42.88 %        *** 2.282991e-05
 url/url-searchparams-sort-rrrr.js n=100000 type="08"    -44.04 %        *** 1.352008e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="09"    -54.74 %        *** 4.523282e-05
 url/url-searchparams-sort-rrrr.js n=100000 type="10"    -49.68 %        *** 8.312344e-09
 url/url-searchparams-sort-rrrr.js n=100000 type="12"    -54.26 %        *** 3.827682e-15
 url/url-searchparams-sort-rrrr.js n=100000 type="14"    -46.84 %        *** 7.272478e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="16"    -31.54 %        *** 1.889203e-05
 url/url-searchparams-sort-rrrr.js n=100000 type="18"    -27.55 %        *** 2.764037e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="20"    -26.76 %        *** 4.446811e-05
 url/url-searchparams-sort-rrrr.js n=100000 type="22"    -25.42 %        *** 9.297027e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="24"    -24.46 %         ** 1.730617e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="26"    -16.97 %         ** 2.375029e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="28"    -19.62 %        *** 4.361287e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="30"    -18.92 %          * 2.231800e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="32"    -10.70 %        *** 3.871433e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="34"    -23.76 %        *** 2.304507e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="36"     -5.79 %            2.411327e-01
 url/url-searchparams-sort-rrrr.js n=100000 type="38"    -20.30 %         ** 7.036384e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="40"    -10.63 %            7.832040e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="42"      6.09 %            1.150079e-01
 url/url-searchparams-sort-rrrr.js n=100000 type="44"     -1.04 %            7.316006e-01
 url/url-searchparams-sort-rrrr.js n=100000 type="46"    -14.07 %          * 1.335973e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="48"     -6.45 %          * 4.874718e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="50"     11.42 %        *** 7.072396e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="52"      7.85 %          * 1.622392e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="54"     21.93 %         ** 1.438206e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="56"     16.34 %         ** 9.748436e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="58"      6.05 %          * 3.993576e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="60"     28.51 %        *** 1.892919e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="62"     30.69 %         ** 8.003372e-03
 url/url-searchparams-sort-rrrr.js n=100000 type="64"     56.55 %        *** 1.434437e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="66"     17.48 %        *** 1.588903e-04
 url/url-searchparams-sort-rrrr.js n=100000 type="68"     65.50 %        *** 2.107887e-07
 url/url-searchparams-sort-rrrr.js n=100000 type="70"     38.83 %        *** 3.505777e-07
 url/url-searchparams-sort-rrrr.js n=100000 type="72"      5.23 %            7.135306e-02
 url/url-searchparams-sort-rrrr.js n=100000 type="74"     56.92 %        *** 5.389008e-08
 url/url-searchparams-sort-rrrr.js n=100000 type="76"     74.65 %        *** 5.051242e-10
 url/url-searchparams-sort-rrrr.js n=100000 type="78"     24.76 %        *** 3.956218e-06
 url/url-searchparams-sort-rrrr.js n=100000 type="80"     30.39 %        *** 9.959930e-05

TimothyGu · 2017-02-01T23:24:41Z

CI: https://ci.nodejs.org/job/node-test-pull-request/6155/

TimothyGu · 2017-02-03T05:29:39Z

Added Web Platform Tests and rebased.

New CI: https://ci.nodejs.org/job/node-test-pull-request/6186/

jasnell

Almost there.

jasnell · 2017-02-03T17:05:33Z

doc/api/url.md

+params.sort();
+console.log(params.toString());
+  // Prints query%5B%5D=123&query%5B%5D=abc&type=search
+```


It may be better to swap the query[]=123 and query[]=abc positions to show that the sort will not affect the ordering of the abc and 123 positions given that 123 would typically appear before abc

Good idea. Done.

TimothyGu · 2017-02-05T02:17:41Z

@mscdex, have you had a chance yet to look at this? Frankly, I'm more than happy with its performance, as my makeshift insertion/merge sort hybrid is able to outperform Array.prototype.sort() (which isn't even stable).

targos · 2017-02-05T08:00:02Z

lib/internal/url.js

+    } else {
+      // Bottom-up iterative stable merge sort
+      const lBuffer = Array(len);
+      const rBuffer = Array(len);


nit: new Array

I agree with @targos.

joyeecheung · 2017-02-05T09:39:19Z

@TimothyGu There is a cost of memory though, Array.prototype.sort is in-place.

mscdex · 2017-02-05T17:09:25Z

benchmark/url/url-searchparams-sort.js

+  // 8 elements
+  short: 'm&t&d&c&z&v&a&n',
+  // 88 elements
+  long: 'g&r&t&h&s&r&d&w&b&n&h&k&x&m&k&h&o&e&x&c&c&g&e&b&p&p&s&n&j&b&y&z&' +


This won't be 88 parameters because only unique keys are kept by querystring.parse(). One suggestion is to use unique keys first:

long: 'g&r&t&h&s&rr&d&w&b&n&hh&k&x&m&kk&hhh&o&e&xx&c&cc&gg&ee&bb&' + 'p&pp&ss&nn&j&bbb&y&z&u&l&oo&rrr&ww&a&uu&ll&mm&f&jj&q&ppp&ff&' + 'eee&yy&eeee&nnn&eeeee&lll&mmm&www&uuu&wwww&tt&nnnn&ttt&qq&v&' + 'yyy&ccc&ooo&kkk&fff&jjj&i&llll&mmmm&ggg&jjjj&dd&ii&zz&qqq&pppp&' + 'xxx&qqqq&qqqqq&ddd&nnnnn&yyyy&wwwww&gggg&iii&vv&rrrr'

and then add this before the timed loop to create duplicates:

if (conf.type === 'long') { // Make `array` contain duplicate keys, useful for benchmarking stable-ness // of sorting for (i = 0; i < array.length; i += 2) { if (array[i].length > 1) array[i] = array[i][0]; } }

This should trigger the current alternate code path that uses merge sort.

Right, but querystring.parse returns something like { g: [ '', '' ] } in this case, and this object is converted to [ 'g', '', 'g', '' ]. Either way, I added a small ad-hoc parser to skip querystring.parse, and to also maintain the order of the input.

I think that's just a bug in the current querystring.parse() in master. For example, querystring.parse('a&a&a&a') currently returns { a: [ '', '' ] } instead of { a: [ '', '', '', '' ] }. This bug is probably due to 4e259b2 which may/may not be fixed by #11171. At any rate, using unique keys in the beginning will avoid such possible bugs.

mscdex · 2017-02-05T17:10:50Z

benchmark/url/url-searchparams-sort.js

+  almostsorted: 'a&b&c&d&e&f&g&i&h&j&k&l&m&n&o&p&q&r&s&t&u&w&v&x&y&z',
+  reversed: 'z&y&x&w&v&u&t&s&r&q&p&o&n&m&l&k&j&i&h&g&f&e&d&c&b&a',
+  random: 'm&t&d&c&z&v&a&n&p&y&u&o&h&l&f&j&e&q&b&i&s&x&k&w&r&g',
+  // 8 elements


I think a better name here would be 'search parameters' or similar. The term 'elements' can be confusing if you're familiar with the underlying URLSearchParams implementation (e.g. does 'elements' mean number of pairs or params[searchParams].length?).

mscdex · 2017-02-05T17:13:30Z

benchmark/url/url-searchparams-sort.js

+
+const bench = common.createBenchmark(main, {
+  type: Object.keys(inputs),
+  n: [1e5]


This might be a little low IMHO. The results came back awfully fast on my machine, which could mean V8 didn't have enough time to properly optimize functions. For example, I used ~5e6 when testing type=long.

Changed to 1e6.

If the results come back too fast there is a possibility that this got loop-invariant code-motioned. I usually get alarmed when I see op/s come close to the op/s of an empty loop on my machine.

mscdex · 2017-02-05T17:14:01Z

lib/internal/url.js

+        var j;
+        for (j = i - 2; j >= 0; j -= 2) {
+          var tmpKey = a[j];
+          var tmpVal = a[j + 1];


If you remove these two temporary variables (use the values as-is where needed), you will get a small perf boost.

mscdex · 2017-02-05T17:35:40Z

Without diving into the multitude of sort algorithms out there to refresh my memory, these changes are probably fine, except for a few minor nits.

One unrelated nit I noticed is that because the whatwg URL module uses querystring, the order of parameters is not preserved. From what I read the spec does not explicitly state the order in which parameters are returned (for example in params.toString()). At least Chrome preserves the original order (as long as .sort() is not used of course), but node's implementation groups duplicate keys together. I just thought I'd throw that out there.

mscdex · 2017-02-05T17:41:18Z

I should also say I'm a bit confused about the "cutoff" being 59. In the code the conditional is len < 90 which would mean 45 pairs. Either way, 90 or 45 are not 59. It's not obvious which count the "cutoff" is taking into account. @TimothyGu Did you perhaps change the "cutoff" value after your original post?

TimothyGu · 2017-02-05T17:44:25Z

@mscdex, yes I changed the cutoff after moving merge() out of sort(), which helped the performance a bit. Sorry for not making that clear. I'll look at your comments and see if there's anything more I can do.

TimothyGu · 2017-02-05T18:26:40Z

@joyeecheung

There is a cost of memory though, Array.prototype.sort is in-place.

For most of the cases (when there are fewer than 50 queries), the in-place selection sort is used. For >= 50, merge sort at most makes two more copies of the params array, so O(n) space complexity. (It is also the algorithm used by SpiderMonkey's Array.prototype.sort.) At the end of the day, chances are, sort() probably isn't the function using up the most memory in an app either way.

TimothyGu · 2017-02-05T18:27:04Z

@jasnell, @targos, @mscdex, PTAL.

CI: https://ci.nodejs.org/job/node-test-pull-request/6224/

mscdex · 2017-02-05T18:29:14Z

For >= 50, merge sort at most makes two more copies of the params array, so O(n) space complexity.

You mean O(2n)?

TimothyGu · 2017-02-05T18:31:27Z

You mean O(2n)?

Coefficients are irrelevant in Big O notation 😺

mscdex · 2017-02-05T18:33:28Z

benchmark/url/url-searchparams-sort.js

+  const searchParams = require('internal/url').searchParamsSymbol;
+  const input = inputs[conf.type];
+  const n = conf.n | 0;
+  const params = new URLSearchParams;


IMHO () should be added after the constructor name.

TimothyGu · 2017-02-08T07:42:58Z

@jasnell, @targos, @mscdex, can I get some LGTMs please :)

mscdex · 2017-02-08T08:52:11Z

CI: https://ci.nodejs.org/job/node-test-pull-request/6280/

Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

TimothyGu · 2017-02-11T19:47:16Z

Rebased. Going to apply on Monday if there are no more objections.

CI: https://ci.nodejs.org/job/node-test-pull-request/6356/

TimothyGu · 2017-02-14T04:50:29Z

Landed in 02d1e32 and 781eb90.

PR-URL: #11098 Fixes: #10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>

Backport-of: nodejs#11098 Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

Backport-of: nodejs#11098

Backport-of: nodejs#11098 Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

Backport-of: nodejs#11098

Backport-of: #11098 Fixes: #10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

Backport-of: #11098

nodejs-github-bot added dont-land-on-v4.x whatwg-url Issues and PRs related to the WHATWG URL implementation. labels Feb 1, 2017

TimothyGu mentioned this pull request Feb 1, 2017

url: track status of a spec-compliant URLSearchParams #10821

Closed

11 tasks

jasnell requested a review from mscdex February 1, 2017 16:48

jasnell requested changes Feb 1, 2017

View reviewed changes

joyeecheung approved these changes Feb 1, 2017

View reviewed changes

joyeecheung reviewed Feb 1, 2017

View reviewed changes

TimothyGu added the semver-minor PRs that contain new features and should be released in the next minor version. label Feb 3, 2017

TimothyGu force-pushed the urlsearchparams-sort branch from ba709b9 to 7cbbea6 Compare February 3, 2017 05:28

jasnell requested changes Feb 3, 2017

View reviewed changes

targos reviewed Feb 5, 2017

View reviewed changes

mscdex reviewed Feb 5, 2017

View reviewed changes

TimothyGu force-pushed the urlsearchparams-sort branch from 9945032 to c0ea7d0 Compare February 5, 2017 18:27

mscdex reviewed Feb 5, 2017

View reviewed changes

jasnell approved these changes Feb 8, 2017

View reviewed changes

TimothyGu added 2 commits February 11, 2017 11:17

url: add urlSearchParams.sort()

798b358

Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

benchmark: add url/url-searchparams-sort.js

54dde7e

TimothyGu force-pushed the urlsearchparams-sort branch from 2c925d2 to 54dde7e Compare February 11, 2017 19:18

TimothyGu closed this Feb 14, 2017

TimothyGu deleted the urlsearchparams-sort branch February 14, 2017 04:50

TimothyGu added a commit to TimothyGu/node that referenced this pull request Feb 18, 2017

url: add urlSearchParams.sort()

01ac09c

Backport-of: nodejs#11098 Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

TimothyGu added a commit to TimothyGu/node that referenced this pull request Feb 18, 2017

benchmark: add url/url-searchparams-sort.js

18df02a

Backport-of: nodejs#11098

TimothyGu mentioned this pull request Feb 18, 2017

(v7.x backport) URL-related changes #11457

Closed

4 tasks

italoacasas added the dont-land-on-v7.x label Feb 20, 2017

italoacasas pushed a commit to italoacasas/node that referenced this pull request Feb 20, 2017

url: add urlSearchParams.sort()

e634b12

Backport-of: nodejs#11098 Fixes: nodejs#10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

italoacasas pushed a commit to italoacasas/node that referenced this pull request Feb 20, 2017

benchmark: add url/url-searchparams-sort.js

6a8b2cf

Backport-of: nodejs#11098

italoacasas pushed a commit that referenced this pull request Feb 22, 2017

url: add urlSearchParams.sort()

c25c16c

Backport-of: #11098 Fixes: #10760 Ref: whatwg/url#26 Ref: whatwg/url#199 Ref: web-platform-tests/wpt#4531

italoacasas pushed a commit that referenced this pull request Feb 22, 2017

benchmark: add url/url-searchparams-sort.js

7e37628

Backport-of: #11098


		update(this[context], this);

		function merge(out, start, mid, end, lBuffer, rBuffer) {

url: add urlSearchParams.sort() #11098

url: add urlSearchParams.sort() #11098

Conversation

TimothyGu commented Feb 1, 2017 • edited Loading

Checklist

Affected core subsystem(s)

jasnell commented Feb 1, 2017

jasnell left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

joyeecheung left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TimothyGu commented Feb 1, 2017 • edited Loading

TimothyGu commented Feb 1, 2017

TimothyGu commented Feb 1, 2017

TimothyGu commented Feb 3, 2017

jasnell left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TimothyGu commented Feb 5, 2017

Choose a reason for hiding this comment

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

joyeecheung commented Feb 5, 2017

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mscdex Feb 5, 2017 • edited Loading

Choose a reason for hiding this comment

mscdex commented Feb 5, 2017 • edited Loading

mscdex commented Feb 5, 2017

TimothyGu commented Feb 5, 2017

TimothyGu commented Feb 5, 2017

TimothyGu commented Feb 5, 2017 • edited Loading

mscdex commented Feb 5, 2017 • edited Loading

TimothyGu commented Feb 5, 2017

Choose a reason for hiding this comment

TimothyGu commented Feb 8, 2017

mscdex commented Feb 8, 2017

TimothyGu commented Feb 11, 2017

TimothyGu commented Feb 14, 2017

TimothyGu commented Feb 1, 2017 •

edited

Loading

TimothyGu commented Feb 1, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex Feb 5, 2017 •

edited

Loading

mscdex commented Feb 5, 2017 •

edited

Loading

TimothyGu commented Feb 5, 2017 •

edited

Loading

mscdex commented Feb 5, 2017 •

edited

Loading