Issue #21: Report Component #48

Luolc · 2017-07-05T14:01:14Z

I decide to push the changes step by step and confirm the unclear things in the meantime.

This is not a completed PR currently.

As mentioned at #21 (comment), I wrote some code to download the script from remote repo. And actually it is not so different with putting it in our project directly. I assume we would only need diff.groovy in the contribution repo(and list of repos to test on), is it correct?

I think I need more discussion about this component. And here is the simple script downloading code, in case you have any suggestions about this.

rnveach · 2017-07-11T10:48:06Z

I wrote some code to download the script from remote rep

I think we may need to take the contribution repo as an argument in Main.
The groovy scripts still call the contribution's pom so we can't just copy the scripts alone. The scripts also store repositories and such in relative directories and the output of the reports.
I originally wanted to move them here, but I think that may not be as easy.
For now, let's take the easy route and require it as an argument our program.
Sound good?

rnveach · 2017-07-12T12:51:38Z

@Luolc ping
CI is failing.
If you agree as argument to main, don't forget to add it to main issue.

Luolc · 2017-07-14T11:14:24Z

Some discussion on Gitter:

I think all the problems of running the groovy script is that I am impossible to change the working directory of the invoker. There are many relative path used in the script, and we would have problems when invoking it outside its parent folder. Even in CLI, we are not able to run it outside. I am using CLI version now and I think we may consider updating diff.groovy to fix this in the future.

Invoking script with GroovyShell or smth else is impossible now. We could only use CLI way until we update diff.groovy and launch.groovy. CLI version is pushed.

We mentioned deployment should be in this component, and are we going to use the build-in deploy function in Travis currently? If yes, then this should be done by changes of .travis.yml

rnveach · 2017-07-19T08:47:14Z

I think we may need to take the contribution repo as an argument in Main.
CLI version is pushed.

Please mention if you added something to issue like new argument. Since you edited, I didn't get a notification of the change.

We mentioned deployment should be in this component

This will be a separate PR and we should only worry about this later.

are we going to use the build-in deploy function in Travis currently? If yes, then this should be done by changes of .travis.yml

Please explain this as I am not sure what you mean by "built-in deploy".
If we want users to be able to deploy reports locally, they won't have travis built-in.

IMO, we should probably have different methods for deploy taken as arguments.
Ex: Github push, no deploy, etc...

rnveach · 2017-07-19T08:42:02Z

pom.xml

+                                    <regex>
+                                        <pattern>com.github.checkstyle.regression.report.ReportGenerator</pattern>
+                                        <branchRate>0</branchRate>
+                                        <lineRate>0</lineRate>


Needs issue.

rnveach · 2017-07-19T08:49:01Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+     * @throws IOException          failure of running CLI
+     */
+    public static File generate(
+            String testerPath, String repoPath, String branch, String configPath)


https://github.com/checkstyle/regression-tool/blob/master/src/main/java/com/github/checkstyle/regression/configuration/ConfigGenerator.java#L126
Shouldn't configPath be a file?

I am unsure of other paths since we don't have code in Main. They should be converted to a File at some point to test if they exist and such.

I am OK with a File parameter. Is there any possibility that user might use his own config file and don't use the generated one?

Is there any possibility that user might use his own config file and don't use the generated one?

It is a possibility, but like I stated, anything that comes from user has to converted to a File at some point to test if it exists and such. We should probably do this sooner rather than later.

Updated. It is now a File type. And I add a check of the existence. Do you think it is proper to check here, or in Main, before passing it to the method?

Main should do all validation to check if inputs are correct.

rnveach · 2017-07-19T08:49:37Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+                )
+                .inheritIO();
+        final Process process = builder.start();
+        final int code = process.waitFor();


Will users' see output of groovy command?

Yes. inheritIO() makes users see output of the CLI. We could remove inheritIO() if we don't want user see the output.

They need to see output as any error will be in groovy and not displayed by our process.
This is why I wanted to confirm.

rnveach · 2017-07-19T08:49:57Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+        final Process process = builder.start();
+        final int code = process.waitFor();
+        if (code != 0) {
+            throw new IllegalStateException("an error occurs when running diff.groovy");


occurs => occurred

rnveach · 2017-07-19T08:52:53Z

config/pmd.xml

+  <rule ref="rulesets/java/controversial.xml/UseObjectForClearerAPI">
+    <properties>
+      <!-- Considering fix this later, since the interface might be changed. -->
+      <property name="violationSuppressXPath" value="//ClassOrInterfaceDeclaration[@Image='ReportGenerator']"/>


Make issue.
Or if will be possibly fixed in this issue,make note of it in issue.

This is removed after we changed parameter configPath to a File type.

rnveach · 2017-07-19T12:19:50Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+            String testerPath, String repoPath, String branch, File configFile)
+            throws InterruptedException, IOException {
+        if (!configFile.exists()) {
+            throw new IllegalArgumentException("config file does not exist: " + configFile);


Like I said, Main should do all validation and testing to check if inputs are valid.
All components should assume it is given valid information.

Please remove this if check.

rnveach · 2017-07-19T12:20:25Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+        if (code != 0) {
+            throw new IllegalStateException("an error occurred when running diff.groovy");
+        }
+        return new File(testerPath, "reports/diff");


Should we test if it exists and is directory before continuing? We don't want to return bad data.

rnveach · 2017-07-19T12:28:46Z

src/main/java/com/github/checkstyle/regression/report/ReportGenerator.java

+        final Process process = new ProcessBuilder()
+                .directory(new File(testerPath))
+                .command(
+                        "groovy", "diff.groovy",


One final thing, let's make a new issue on using GroovyShell.
Please include all details and any supporting documentation or stack overflows.

Luolc force-pushed the issue21 branch 2 times, most recently from da622e2 to 4f558e2 Compare July 11, 2017 04:49

Luolc force-pushed the issue21 branch from 4f558e2 to 6511675 Compare July 11, 2017 13:28

Luolc force-pushed the issue21 branch 2 times, most recently from 98f4538 to 32ac855 Compare July 14, 2017 11:05

rnveach requested changes Jul 19, 2017

View reviewed changes

Luolc force-pushed the issue21 branch from 32ac855 to 5aeb207 Compare July 19, 2017 10:36

Luolc mentioned this pull request Jul 19, 2017

Codecov: ReportGenerator #55

Open

rnveach requested changes Jul 19, 2017

View reviewed changes

rnveach reviewed Jul 19, 2017

View reviewed changes

Issue checkstyle#21: Report Component

62fd7a0

Luolc force-pushed the issue21 branch from 5aeb207 to 62fd7a0 Compare July 19, 2017 15:27

Luolc mentioned this pull request Jul 19, 2017

Report: using GroovyShell instead of running CLI directly #56

Open

rnveach approved these changes Jul 20, 2017

View reviewed changes

rnveach merged commit bc20f0f into checkstyle:master Jul 20, 2017

Luolc deleted the issue21 branch July 20, 2017 12:22

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Issue #21: Report Component #48

Issue #21: Report Component #48

Luolc commented Jul 5, 2017 •

edited

rnveach commented Jul 11, 2017

rnveach commented Jul 12, 2017 •

edited

Luolc commented Jul 14, 2017 •

edited

rnveach commented Jul 19, 2017 •

edited

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017 •

edited

Luolc Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017

Luolc Jul 19, 2017

rnveach Jul 19, 2017 •

edited

Luolc Jul 19, 2017

Issue #21: Report Component #48

Issue #21: Report Component #48

Conversation

Luolc commented Jul 5, 2017 • edited

rnveach commented Jul 11, 2017

rnveach commented Jul 12, 2017 • edited

Luolc commented Jul 14, 2017 • edited

rnveach commented Jul 19, 2017 • edited

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

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

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

rnveach Jul 19, 2017 • edited

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rnveach Jul 19, 2017 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Luolc commented Jul 5, 2017 •

edited

rnveach commented Jul 12, 2017 •

edited

Luolc commented Jul 14, 2017 •

edited

rnveach commented Jul 19, 2017 •

edited

rnveach Jul 19, 2017 •

edited

rnveach Jul 19, 2017 •

edited