Contribute to Open Source. Search issue labels to find the right project for you!

Fix: Search bar svg is not showing in production


<!— Provide a general summary of the issue in the Title above –>

Types of issue

<!— What types of issue are we looking at? Put an x in all the boxes that apply: –> - [ ] Bug (High impact or critical to a production build) - [x] Bug (Low impact or not working as expected) - [ ] New feature (non-breaking change which adds functionality) - [ ] Something else (idea, build tool, documentation improvement)


<!— Describe your issue in detail –> The svg in search bar component is missing

Possible Solution

<!— Not obligatory, but suggest a fix/reason for the bug, –> <!— or ideas how to implement the addition or change. Can you point to similar functionality with any existing libraries or components? –>

Change the path of the svg in search component


<!— A picture speaks a thousand words. Include one to help us see your problem or understand your request –> screen shot 2017-12-15 at 10 17 38 am

Updated 14/12/2017 23:39

Flyout service amputates ids with a hash character "#"


In the suggest API, it seems like the flyout service returns a partial id string if the id has a “#” character. For example, if the id is somedomain#sometopic and the flyout endpoint is /api/flyout, openrefine will try to make a request to


instead of


This can be reproduced in OpenRefine2.8 on Mac OS 10.13

Any idea?

Updated 14/12/2017 23:28 2 Comments

APKs not being generated


Actual Behaviour

Travis build scripts were based on relative paths and changing of directories and hence, after the PR #2013, it can’t reference to built files

Expected Behaviour

Either travis scripts should reference correct relative paths or preferably use absolute paths so that the directory containing scripts do not have any significance in build process


Last build on Travis :

Updated 14/12/2017 12:26 3 Comments

Support `else if (...)`



The lack of else if is awkward, makes it difficult to write some code.

A quick solution might be to encode the node as a literal else block containing a single if.

Sample spec js export function test(x: i32): i32 { if (x == 0) { return 2; } else if (x == 1) { return 4; } else { return -1; } }

That should do it.

  • [ ] Tests
  • [ ] Tests pass
Updated 14/12/2017 13:52

FitFailedWarning raised by cross validation could do with better message


Current message says “Classifier fit failed” but the estimator is not necessarily a classifier. (Perhaps that’s too pedantic of me…)

%r is not technically the best way to display an error message. We could either use traceback.format_exc and include the whole traceback; or we can use traceback.format_exception_only to print it properly (though I think this is the same as "%s: %s" % (type(exc), exc)).

Another thing we can consider, now that _fit_and_score provides structured results to *SearchCV and cross_validate, is to store the full traceback in some array of *SearchCV.cv_results_.

Updated 14/12/2017 20:54 1 Comments

Change the default of *SearchCV's error_score to np.nan


It can be pretty unpleasant when your grid search fails. That’s why the error_score parameter was introduced. The default error_score='raise' helps make absolutely clear that there was a failure by terminating the program, but it might be kind to just set the score to NaN and raise a warning. We could consider deprecating it: * change the default temporarily to error_score='raise-deprecating' * continue to raise an error with 'raise-deprecating' but also warn that in the future, this error will not be raised, and the user can choose between error_score='raise' or the new default, error_score=np.nan. * change the default to error_score=np.nan in a couple of versions time

Other devs may want to give feedback on this idea before we go ahead with implementation/merge.

Updated 15/12/2017 03:51 2 Comments

BUG Inconsistent f1_score behavior when combining label indicator input with labels attribute



When using label indicator inputs for y_pred and y_true, metrics.f1_score calculates the macro average over all label-specific f-scores whenever the labels parameter includes column index 0. It should only average over the label-specific scores indicated by the labels parameter, as it does when 0 is not present in the labels parameter.

<!– Instructions For Filing a Bug: –>

Steps/Code to Reproduce

import numpy as np
from sklearn.metrics import f1_score, precision_recall_fscore_support

y_true = np.array([[0, 1, 0, 0],
                   [1, 0, 0, 0],
                   [1, 0, 0, 0]])
y_pred = np.array([[0, 1, 0, 0],
                   [0, 0, 1, 0],
                   [0, 1, 0, 0]])

p, r, f, s = precision_recall_fscore_support(y_true, y_pred)
print(f1_score(y_true, y_pred, labels=[0,1], average='macro'))
print(f1_score(y_true, y_pred, labels=[0,1,2], average='macro'))
print(f1_score(y_true, y_pred, labels=[1,3], average='macro'))
print(f1_score(y_true, y_pred, labels=[1,2,3], average='macro'))

Expected Results

[ 0.          0.66666667  0.          0.        ]

Actual Results

[ 0.          0.66666667  0.          0.        ]

<!– Please paste or specifically describe the actual output or traceback. –>


Windows-7-6.1.7601-SP1 Python 3.5.3 |Anaconda custom (64-bit)| (default, May 15 2017, 10:43:23) [MSC v.1900 64 bit (AMD64)] NumPy 1.13.1 SciPy 0.19.0 Scikit-Learn 0.19.0

<!– Thanks for contributing! –>

Updated 14/12/2017 20:53 3 Comments

Make Span IndexOfAny a generic method with special case for byte


Similar to LastIndexOfAny (added here -, change the IndexOfAny APIs to be generic. The premise is that there should be little performance degradation for the byte case when we make this change. This needs to be confirmed by comparing performance results before/after.

Change: ```C# public static int IndexOfAny(this Span<byte> span, byte value0, byte value1) { throw null; } public static int IndexOfAny(this Span<byte> span, byte value0, byte value1, byte value2) { throw null; } public static int IndexOfAny(this Span<byte> span, ReadOnlySpan<byte> values) { throw null; }

public static int IndexOfAny(this ReadOnlySpan<byte> span, byte value0, byte value1) { throw null; } public static int IndexOfAny(this ReadOnlySpan<byte> span, byte value0, byte value1, byte value2) { throw null; } public static int IndexOfAny(this ReadOnlySpan<byte> span, ReadOnlySpan<byte> values) { throw null; } ```

To: ```C# public static int IndexOfAny<T>(this Span<T> span, T value0, T value1) where T : IEquatable<T> { throw null; } public static int IndexOfAny<T>(this Span<T> span, T value0, T value1, T value2) where T : IEquatable<T> { throw null; } public static int IndexOfAny<T>(this Span<T> span, ReadOnlySpan<T> values) where T : IEquatable<T> { throw null; }

public static int IndexOfAny<T>(this ReadOnlySpan<T> span, T value0, T value1) where T : IEquatable<T> { throw null; } public static int IndexOfAny<T>(this ReadOnlySpan<T> span, T value0, T value1, T value2) where T : IEquatable<T> { throw null; } public static int IndexOfAny<T>(this ReadOnlySpan<T> span, ReadOnlySpan<T> values) where T : IEquatable<T> { throw null; } ```

Example implementation change: C# public static int IndexOfAny<T>(this Span<T> span, T value0, T value1) where T : IEquatable<T> { if (typeof(T) == typeof(byte)) return SpanHelpers.IndexOfAny( ref Unsafe.As<T, byte>(ref span.DangerousGetPinnableReference()), Unsafe.As<T, byte>(ref value0), Unsafe.As<T, byte>(ref value1), span.Length); return SpanHelpers.IndexOfAny(ref span.DangerousGetPinnableReference(), value0, value1, span.Length); }

Then implement the following to SpanHelpers.T.cs C# public static unsafe int IndexOfAny<T>(ref T searchSpace, T value0, T value1, int length) where T : IEquatable<T>

Should be relatively easy. cc @karelz

cc @KrzysztofCwalina, @GrabYourPitchforks

Updated 13/12/2017 19:53

Cleaning Up Code


All Open Source Code should be written properly so that it’s easy for other developers to follow. Edit the code to follow the latest JAVA Conventions. This includes proper indentation, correct naming of variables,functions etc. Multiple devs can work on this, and can submit multiple PR’s. We will merge them as long as there is significant improvement in the code.

Updated 13/12/2017 16:24

Não permitir importar sem escolher ao menos um repositório


Comportamento esperado

Não deve-se permitir clicar no botão de importar repositório sem ter escolhido ao menos 1 repositório para importar.

Comportamento atual

É possível clicar no botão de importar sem ter escolhido nenhum respositório.

Passos para reproduzir o comportamento

  • Clique em Import GitHub Repository.
  • Não selecione nenhum repositório na modal.
  • Clique em Import.



  • [ ] A issue possui Labels.
  • [ ] A issue possui prints de tela quando necessário.
  • [ ] A issue possui nome significativo.

**Exclua caso a issue seja um bug

*Exclua caso a issue seja uma história

Updated 13/12/2017 10:44

Connection management


Add back the ability to send another identifier to use as a connection instead of connection. Rethink the connection management story, to support using multiple backends (postgresql, sqlite) in the same code.

Updated 12/12/2017 14:25

Add support for more geoip lookup services


To make OONI Probe more stealth it would be best to not always use the same GeoIP lookup service, but rather use a variety of different strategies to be a bit less fingerprintable.

Currently we only use (see:

In looking at the CISCO umbrella data, it seems like this service it pretty popular (It ranks as number 68980), but not as popular as others.

Here are some other pretty popular ones (I didn’t really do accurate mining of the above mentioned data, but just did some random grepping for common keyword, geoip, ip, lookup etc., probably there is more gold in there):

  • 1619, (
  • 4027,
  • 4767,
  • 15415,
  • 16378,
  • 29838, ( only returns country code)
  • 36926, (only returns country)
  • 37913, (only returns country)
  • 38429,
  • 70241,

The task in this ticket would be to add support for doing geoip lookup via one of any of the above services and maybe open the way for making it more modular.

I think this would be a good “first task”.

Updated 12/12/2017 11:25

linear_model.RidgeClassifierCV's Parameter store_cv_values issue



Parameter store_cv_values error on sklearn.linear_model.RidgeClassifierCV

Steps/Code to Reproduce

import numpy as np from sklearn import linear_model as lm

test database

n = 100 x = np.random.randn(n, 30) y = np.random.normal(size = n)

rr = lm.RidgeClassifierCV(alphas = np.arange(0.1, 1000, 0.1), normalize = True, store_cv_values = True).fit(x, y)

Expected Results

Expected to get the usual ridge regression model output, keeping the cross validation predictions as attribute.

Actual Results

TypeError: init() got an unexpected keyword argument ‘store_cv_values’

lm.RidgeClassifierCV actually has no parameter store_cv_values, even though some attributes depends on it.


Windows-10-10.0.14393-SP0 Python 3.6.3 |Anaconda, Inc.| (default, Oct 15 2017, 03:27:45) [MSC v.1900 64 bit (AMD64)] NumPy 1.13.3 SciPy 0.19.1 Scikit-Learn 0.19.1

Updated 13/12/2017 00:03 5 Comments



Seems like a very Basic gist host that you only access from the command line / telnet / netcat. Raw URLs in the form$LETTERS, and no HTML URLs.

Updated 11/12/2017 15:29

Document edge cases



  • Adblock may be wrong if fingerprint is computed directly after/on page load. Give the browser a few ms to load the adblock plugin (with setTimeout). (What does “a few ms” mean? Does anyone have any experience with concrete values of “a few” here?)
  • Canvas FP may be different on page reload/same browser, same device due to switching between on-chip GPU and “real” GPU. Only seen on Macbook Pro so far. See #147, #280, #226
Updated 11/12/2017 11:59

generate_series for timestamp types


Is this a question, feature request, or bug report?

This is a feature request


Have you checked our documentation at If you could not find an answer there, please consider asking your question in our community forum at, as it would benefit other members of our community.

Prefer live chat? Message our engineers on our Gitter channel at

Tried but, gitter was inaccessible.


  1. Does an issue already exist addressing this request? If yes, please add a :+1: reaction to the existing issue. If not, move on to step 2.

  2. Please describe the feature you are requesting, as well as your proposed use case for this feature.

generate_series(start, stop, step) with timestamp or timestamp with time zone for start and stop parameters and for step interval. Example: SELECT generate_series('2017-11-11 00:00:00'::TIMESTAMP, '2017-11-11 03:00:00'::TIMESTAMP, '1 hour'); will produce: '2017-11-11 00:00:00' '2017-11-11 01:00:00' '2017-11-11 02:00:00' '2017-11-11 03:00:00'

  1. Indicate the importance of this issue to you (blocker, must-have, should-have, nice-to-have). Are you currently using any workarounds to address this issue?

nice-to-have. No feasible workarounds.


  1. Please supply the header (i.e. the first few lines) of your most recent log file for each node in your cluster. On most unix-based systems running with defaults, this boils down to the output of

    grep -F ‘[config]’ cockroach-data/logs/cockroach.log

    When log files are not available, supply the output of cockroach version and all flags/environment variables passed to cockroach start instead.

  2. Please describe the issue you observed:

  3. What did you do?

  4. What did you expect to see?

  5. What did you see instead?

Updated 11/12/2017 17:03 2 Comments

Decrease paste time by inserting line breaks


When pasting bitshuffle-encoded data, large files can take a long time to paste, since the application must read the text into memory. This is exacerbated by large line lengths.

The current standard is to add one line break after each packet. However, with large chuck sizes (>~1000) this decreases load times significantly.

We should add an option, enabled by default, to insert line breaks approximately every 100 characters.

Updated 10/12/2017 23:16 1 Comments

Saving filters under the same name


Saving a filter setting under the same name as an already existing filter adds it to the filter list. Both filters have the same name but only the settings of the previously existing filter are present for both entries. Behavior should be when trying to save a filter with the same name as an existing filter either: a) force saving filter setting under a different name b) overwrite existing filter settings with new settings after user warning c) include prompt with option to do a) or b)

Running v3.3.2 with Android 7.1.1

Updated 09/12/2017 14:09 1 Comments



Issue is only used for submiting bug report and documents typo. If there are same issues or answers can be found in documents, we will close it directly. (为了节约时间,提高处理问题的效率,不按照格式填写的 issue 将会直接关闭。)

Use the commands below to provide key information from your environment: You do NOT have to include this information if this is a FEATURE REQUEST

What version of frp are you using (./frpc -v or ./frps -v)? 0.14.0

What operating system and processor architecture are you using (go env)? ubuntu 14.04 windows 10 64bit

Configures you used: 服务端配置(frps.ini): bind_port = 9100 kcp_bind_port = 9101 privilege_token = 654321

客户端配置(frps.ini): [common] server_addr = x.x.x.x server_port = 9100 protocol = kcp privilege_token = 654321

[vnc] type = tcp local_ip = local_port = 5900 remote_port = 5900 use_encryption = true use_compression = true

Steps to reproduce the issue: 在bind_port 和kcp_bind_port 不一致时(如9100和9101),kcp模式无法连接, 服务端提示: 2017/12/08 12:36:17 [I] [service.go:87] frps tcp listen on 2017/12/08 12:36:17 [I] [service.go:96] frps kcp listen on udp 2017/12/08 12:36:17 [I] [main.go:112] Start frps success 2017/12/08 12:36:17 [I] [main.go:114] PrivilegeMode is enabled, you should pay more attention to security issues 客户端提示: 2017/12/08 20:40:38 [W] [control.go:483] [f24ba18323dbdd2a] reconnect to server error: i/o timeout


Updated 10/12/2017 17:35 3 Comments

"Unrecognized arguments" in CLI Example


From the example in the ‘confidant help’ command, this:

confidant get_service -u "" -k "alias/authnz-production" --from myservice-production --to confidant-production --user_type service --region us-west-2 --service myservice-production

yields the error

confidant: error: unrecognized arguments: -u -k alias/authnz-production --from myservice-production --to confidant-production --user_type service --region us-west-2

It should yield a url unreachable or kms auth error, not an argument error if the example is correct.

Updated 08/12/2017 00:37 5 Comments

uc_kvm_loader.asm refers to bad MIT LICENSE file


It would be best if this file could be relicensed to GPL to avoid having to ship with a copy of the MIT license (CC @phire ). As it stands, OpenSWE1R does not ship with a copy of the MIT license, making this a copyright violation.

Including the MIT or getting permission might not even be necessary though because the file is substentially different from the original file. It essentially has been rewritten entirely. Chances are we can just relicense it to GPL.

Eitherway: This requires changes or the file should be removed. Also related to #9 as we should get rid of this bios anyway.

Updated 12/12/2017 02:59 1 Comments

Peer banning should be temporary


The role of banning a misbehaving peer is preventing repeated offenses, whether it’s DoS or feeding us false information that can take some time to recover from. It mitigates the damaging effect of someone trying to be nefarious or just running buggy code. However we may not want to ban forever.

First, someone who’s nefarious at first may not always be. Second, if you’re running experimental code, being banned forever for a bug sucks. Third, at this point our code isn’t bug-free either. So letting a banned peer connection again after a few hour may be the best approach.

Updated 08/12/2017 22:45 3 Comments

Debug version: SAM output of CRAM file is populated with debug on pipe


The SAM output of a CRAM file is unintentionally populated with debug lines.

  • version of sambamba
sambamba 0.6.7

This version was built with:
    LDC 1.1.1
    using DMD v2.071.2
    using LLVM 3.8.1
    bootstrapped with LDC - the LLVM D compiler (0.17.4)
  • version of samtools that did bam2cram
samtools 1.6
Using htslib 1.6
Copyright (C) 2017 Genome Research Ltd.

System version: sh uname --kernel-release && cat /etc/system-release 2.6.32-696.3.1.el6.x86_64 CentOS release 6.9 (Final)

  • the command line or bash script used (feel free to shorten paths and filenames)
htsfile sampleX.aln.cram
sampleX.aln.cram:   CRAM version 3.0 compressed sequence data
sambamba view -C -h sampleX.aln.cram 2>/dev/null | less -S

Would return SAM format on stdout:

  1. the header
  2. debug lines
  1. alignments

Looking at stderr:

sambamba view -C -h sampleX.aln.cram 2>&1 1>/dev/null | less

The output is:

Init cram_fd* #1
Init cram_fd* #2
Init _Anonymous_25* #1
cram_read_slice (1/1)
Init cram_slice* #1
Init _Anonymous_25* #2
cram_read_slice (1/1)
Init cram_slice* #2
Init _Anonymous_25* #3
cram_read_slice (1/1)
Init cram_slice* #69
Init _Anonymous_5* #4
Free _Anonymous_5* #3
Free cram_slice* #3
Free _Anonymous_25* #3

The debug line on stdout is written on this line: cram/wrappers.d#L162

Problem occurs in both Zsh and Bash.

Updated 07/12/2017 06:40 3 Comments

Unit Tests for API endpoints


The recent update to API endpoints here appear to have broken wallet restore; that’s completely forgivable. What isn’t forgivable is that something like this should have been easily caught in automated testing, which I don’t think we do at all for the API (unless I’m mistaken) or if we do anywhere, they’re very out of date.

Unit tests should be put into place that cover the API endpoints. Should be simple enough, build up a small chain and a wallet in a cargo test, then put a few tests together to make sure all the endpoints return more or less what’s expected.

Updated 13/12/2017 09:57 11 Comments

"Animate" the dice!


This is an easy, but fun upgrade to the current project.

We want to give the user the illusion of dice rolling when they press a key to roll the dice:

How to solve: Look into The dice are stored into an array. How could we iterate through the dice randomly while taking 1-2 seconds to do so before the user sees the result?

Hints: Use the random library already imported to randomly iterate through the different states of the array. import time to access the sleep() method. Roll your new process into a function.

Updated 11/12/2017 18:27 5 Comments

Add QuantileTransformer to PowerTransformer example


I think it would be nice to add the QuantileTransformer to the new PowerTransformer example. I have to say I’m not entirely convinced by which is currently the only comparison we have. The PowerTransformer example also has multi-modal data, which I think is interesting.

On the other hand, is interesting because it shows how the marginals ignore the interdependence of the joint, though I feel that could be made even more explicit with a more complex interdependence.

Anyway, as a first step I’d like to see the QuantileTransformer added to the PowerTransformer example.

Updated 15/12/2017 03:45 13 Comments



We can add some common queries and their solutions here. Be it for a developer or a user.

A very rudimentary example could be adding instructions for squashing commits!

Updated 08/12/2017 08:26 1 Comments

Fork me on GitHub