weixin_39600616的博客

Hi, thanks for your answer, I have tried using double, it is more robust but it doesn't solve, I will prepare a minimal example so the error will be further documented. I can confirm that my net works using other configurations and activation functions.

Moreover, I must say that even with the same data I can run the training several times and I do not always get nan, and I also get different performances (between 80% and 95%). I believe that the variations in the performances depends on the random weights initialization, it should converge increasing epochs and number of neurons. When it goes to 50% it just means that it is not working as it picks classes randomly.

However, what I think is that there should be some nan check in the activation function to detect when a neuron is failing in order to give an exception because that neuron cannot be activated anymore and the error will propagate. This can be an improvement in the robustness of the library.

I faced the issue when upgrading to v0.12.2 from v0.11.3. I guess it's a bug that assume Nilable Marshaler will have Elem at the same time. The following patch can work around, but I',m not sure any other edge cases will happen.

diff
                {{- else if eq ($type.GO | ref) "map[string]interface{}" }}
                    return v.(map[string]interface{}), nil
                {{- else if $type.IsMarshaler }}
-                   {{- if $type.IsNilable }}
+                   {{- if and $type.IsNilable $type.Elem }}
                        var res = new({{ $type.Elem.GO | ref }})
                    {{- else}}
                        var res {{ $type.GO | ref }}
                    {{- end }}
                    err := res.UnmarshalGQL(v)
                    return res, graphql.WrapErrorWithInputPath(ctx, err)

Support slave and master mode runtime config.

Support slave and master mode runtime config.

Added PR as issue #74

I think so. I'm not sure how to ignore, though. This is the code in railties/lib/rails/application.rb that tries to load the config (and throws the error)

def env_config
   ||= begin
    if config.secret_key_base.nil?
      ActiveSupport::Deprecation.warn "You didn't set config.secret_key_base in config/initializers/secret_token.rb file. " +
        "This should be used instead of the old deprecated config.secret_token in order to use the new EncryptedCookieStore. " +
        "To convert safely to the encrypted store (without losing existing cookies and sessions), see http://guides.rubyonrails.org/upgrading_ruby_on_rails.html#action-pack"

      if config.secret_token.blank?
        raise "You must set config.secret_key_base in your app's config"
      end
    end
  ...

I don't know how to bypass that code from rails-api.

We could put back the removed session middleware but condition its use by the api_only flag, as it is done for the Rails 3 stack. Combined with setting api_only to false could fix problem. (Just guessing, haven't tested it).

Are you using an SPA framework or not?

I run this dotnet electronize build /target win7-x86 and everything seems to work fine. Use the code for your publishing or refer to publishing targets in the documentation

Can you share the code or develop a simple sample that produces the problem. If I have been doing my correctly, your output should not show at all.

To create an installer is a bit tricky. The shortest way for me is to rather package the ../obj/desktop/win7-x86 folder with electron-packager if you build for win32. It is easier changing the name of the application, setting the icon and providing other useful information.

While i have not tried that, it should be easier to set the update features from there

The problem is your .csproj. I ran into the same issue today. I had inadvertently modified a part of my csproj i couldn't remember.

Try copying and pasting a working csproj of a different project that works. If the need be, reinstall all nuget packages that will be missing.

Then lastly, avoid updating the netcore packages, e.g. entityframework etc. This project only works with netcore 2 and updating any dependency to a different version can be fatal

In Rails 4 StrongParameters is in the list of modules included by ActionController::Base. In rails-api it's not. I was tempted to add StrongParameters to the list of modules for ActionController::API but that would probably be a bad idea if used for a Rails 3 API app.

Oo my bad. Thanks a ton bro.

Thanks in advance bro, but I think I have properly initialized. Have a look at the complete code. `public class LaunchController implements Initializable {

private JFXTextField searchfield;

private JFXHamburger hammenu;

private JFXListView<Label> view;

private JFXButton addtogroup;

private JFXButton delete;

private JFXButton export;

private JFXButton search;

private JFXButton add;

private AnchorPane pane;

private StackPane stackpane;
private JFXPopup popup;

databasehandler databaseHandler1;






public  void initialize(URL url, ResourceBundle rb) {
    try {
        databaseHandler1 = databasehandler.getInstance();  //getting an object 
        String qu = "Select name FROM CONTACT";
        ResultSet rs = databaseHandler1.execQuery(qu);
        if (rs != null) {
            while (rs.next()) {
                String name = rs.getString("name");
                Label lbl = new Label("     " + (name));
                if (new File("D:\\Work\\Programming\\javaFx Projects\\Contact\\src\\contact\\ui\\files\\Profile\\" + name + ".jpg").exists()) {
                    FileInputStream fis = new FileInputStream("D:\\Work\\Programming\\javaFx Projects\\Contact\\src\\contact\\ui\\files\\Profile\\" + name + ".jpg");
                    Image image = new Image(fis, 35, 35, true, true);
                    lbl.setGraphic(new ImageView(image));
                } else {
                    FileInputStream fis = new FileInputStream("D:\\Work\\Programming\\javaFx Projects\\Contact\\src\\contact\\ui\\files\\person.png");
                    Image image = new Image(fis, 35, 35, true, true);
                    lbl.setGraphic(new ImageView(image));
                }

                view.getItems().add(lbl);
            }
        }
    } catch (FileNotFoundException | SQLException ex) {
        Logger.getLogger(LaunchController.class.getName()).log(Level.SEVERE, null, ex);
    }
   initpopup();
}

private void addnewview(MouseEvent event) throws IOException {
    AnchorPane newpane = FXMLLoader.load(getClass().getResource("/contact/ui/addnew/Add new.fxml"));
    pane.getChildren().setAll(newpane);
}

 private void initpopup(){

// popup initialize
    JFXButton b1 = new JFXButton("View");
    JFXButton b2 = new JFXButton("Edit");
    JFXButton b3 = new JFXButton("Delete");
    JFXButton b4 = new JFXButton("Add to");

    b1.setPadding(new Insets(10));
    b2.setPadding(new Insets(10));
    b3.setPadding(new Insets(10));
    b4.setPadding(new Insets(10));

    VBox  vbox=new VBox(b1,b2,b3,b4);

    popup.setPopupContent(vbox);

 }




private void showPopup(MouseEvent event) {

    if (event.getButton() == MouseButton.SECONDARY) {
        popup.setAutoFix(true);
        popup.show(view,JFXPopup.PopupVPosition.TOP, JFXPopup.PopupHPosition.LEFT, event.getX(), event.getY());
    }

}

} `

Ignore the popup. line.

There is a guide to using a CMK for AMI encryption here: https://aws.amazon.com/blogs/security/how-to-create-a-custom-ami-with-encrypted-amazon-ebs-snapshots-and-share-it-with-other-accounts-and-regions/

This IAM user or role also needs to be able to perform the KMS DescribeKey, CreateGrant, GenerateDataKey and ReEncypt operations on the source account’s KMS CMK, cmkSource.

All of the above operations are supported by aws_kms_grant, note I'm not sure if you need both ReEncryptTo and ReEncryptFrom.

(As a side note, RevokeGrant requires special privileges see the docs, and RetireGrant has the exact same functionality. So if you ever need it in the future you should probably use RetireGrant).

You can see the list of valid operations here: https://docs.aws.amazon.com/cli/latest/reference/kms/create-grant.html

Where valid values are: Decrypt Encrypt GenerateDataKey GenerateDataKeyWithoutPlaintext ReEncryptFrom ReEncryptTo CreateGrant RetireGrant DescribeKey

I will work on this.

looks like the issue has been fixed in #1511

Having the same issue as . I have already set hrefTargetBlank = true. Everything works fine on my localhost:1313 but on the published website (hosted by Netlify) it's error behavior: external site opened in original AND new tab.

Edit: Changing it to false didn't change any behavior either. Did I miss something?

Why not use

use MongoDB\BSON\UTCDateTime;

new UTCDateTime('your date');

$modelResult = [
    'date'=> $obj['date']->toDateTime()
]

Same request - the source code hard is to find otherwise.

+1. Rolled back to Laravel v5.3.18, which works.

Many thanks - that worked a treat. It was in my own code, and now that I see where the problem was, I shouldn't have assumed a blanket search over my own project for the word "title" wouldn't have worked... But this change you've implemented should help other folks experiencing deprecation warnings in the future.

Thanks for getting back to me. Here goes the gemfile - I am using RefineryCMS and am trying to upgrade to version 3. Had no issues before the upgrade - but the upgrade has affected quite a few gems. I am getting no stack trace for deprecation warnings. Please let me know how I can switch it on for cucumber.

source 'https://rubygems.org'
ruby "2.1.3"

gem 'rails'
gem 'pg'
gem 'fog'
gem 'delayed_job_active_record' #, "=0.3.3"  # without PG 9.1 you get lock errors
gem "workless", "~> 1.1.3"
gem "daemons"
gem 'jquery-rails'
gem 'carmen-rails', '~> 1.0.0'
gem 'gibbon' #, '~> 0.3.5'
gem 'memcachier'
gem 'dalli'
gem 'twitter'
gem 'sass', '~> 3.2.15'  # as per https://github.com/sass/sass/issues/1162, errors while saving cache at Heroku, can't dump anonymous class
gem "acts-as-taggable-on"  # needed for rake tasks for refinerycms-blog

group :development, :test do
  gem 'yaml_db', :git => "git://github.com/lostapathy/yaml_db.git"
  gem 'refinerycms-testing' #, '~> 2.1.2'
end

group :test do
  gem 'cucumber' 
  gem 'cucumber-rails', require: false # required code for refinerycms 2.1.2 upgrade
  gem 'capybara-webkit'
  gem 'launchy'
  gem 'turn', :require => false
  gem 'minitest'
  gem 'email_spec'
  gem 'rspec-activemodel-mocks'
end

group :production, :staging do
  gem 'newrelic_rpm'
  gem 'rails_12factor'
  gem 'heroku-deflater', '0.5.2'  # enables gzip compression for rails on heroku, locked to 0.5.2 because of https://github.com/romanbsd/heroku-deflater/issues/15
  gem "dragonfly-s3_data_store"
end

gem 'unicorn'

# Refinery CMS
gem 'refinerycms', '~>3.0.0'

gem 'refinerycms-blog', :git => 'git://github.com/refinery/refinerycms-blog.git', :branch => 'master'
gem 'refinerycms-inquiries', :git => 'git://github.com/refinery/refinerycms-inquiries.git', :branch => 'master'
gem 'refinerycms-authentication-devise', '~> 1.0'

gem 'refinerycms-prayer_points', :path => 'vendor/extensions'
gem 'refinerycms-tours', :path => 'vendor/extensions'

ALL_BUILD succeeds with CMake GUI

thanks , let us know if u need any help w/ Travis CI

OK, so I am not closing this. can you flag this as a separate Win issue?

之前测试过，可以防撤回，但是不知道什么时候失效了，难道是因为更新的问题？

I encountered the same problem when trying to convert a customized PyTorch model. Have you solved this problem?

It works! Thanks!

Thanks, I think my problem is mostly related to #281 but it has not been solved in #281 yet. Do you know why torch.load() cannot load a modified class?

thanks for the review. I see some checks failed but I can't see the reasoning. Please let me know if it is anything on my side

thanks for the review. I made a change to the PR. Please verify that the changes made are what your thoughts were.

A better solution might be the following:

https://github.com/mongodb/mongo-go-driver/blob/4503c54ec045b406a0b3533901c6d7a5614cfef3/mongo/clientopt/clientopt.go#L557

if !c.ConnString.SSLClientCertificateKeyFileSet {
   if ssl.ClientCertificateKeyFile != "" {
      c.ConnString.SSLClientCertificateKeyFile = ssl.ClientCertificateKeyFile
      c.ConnString.SSLClientCertificateKeyFileSet = true 
   }
}

Looking for feedback from the dev team.

I'm seeing this security rule duplicate error somewhat regularly about every other day, I checked with AWS and it seems that terraform is indeed trying to create some duplicate security group rules.

As people have been that have been in the security group rule resource file recently, do you know what might be causing this? I'm having some trouble diagnosing the issue, if you have any hints I'd be happy to investigate further and submit a PR fixing the issue.

I'm also seeing this error on:

• terraform v0.11.7
• terraform provider aws v1.16.0

I've been running into the same issue recently as well. Noticed, the lingering ENIs are almost always previously attached to an ELB. Are you still running into the issue?

Sorry my poor writing at 1am was to do with malsups attempt at recreation of the issue. The server your sending your () request too would be redirecting also back to your server(or whatever nexturl is). So this issue involves CORS, uploadProgress Handler, and Redirection of CORS request as CORS seems to work without redirection. Its almost as if the xhr forgets how to redirect if you add an uploadProgress handler. ?

I am chrome: Version 25.0.1364.172 m on Windows 7 64

If it helps.. adding an progress handler makes the cors do a preceeding options request (which passes) where without progress handler it doesn't.

Oh after looking at your(malsup's) code to reproduce you should add the 303 redirect as response as that is also in the mix of the problem.

I am also having this issue. Using Fiddler I can see that the server is sending a redirect response but the request maintains a pending type and (eventually) cancelled status. The chrome client never sees a response. It eventually turns from pending to cancelled. I also have narrowed it down 100% to adding the handler for uploadProgress as adding/removing the handler will add/remove the issue. I also get a good progress reporting. The response from the server is a 303 with new location(back to my server). Works fine in IE10? (wtf when does that ever happen). File is received fine by S3 (the CORS server).

I can confirm that current master fixes the bug. Thanks!

Assuming we can get AWS to commit to not changing the structure of the pagination token/marker (I can sort that out), would a PR implementing the homegrown DescribeGrant call above be merged?

This is definitely worth implementing if AWS can commit to making the pagination token/marker stable.

Thank you for all the help and comments. I updated the documentation, and replied to your comments. Should be ready to merge. Let me know if you want me to squash the commits.