GeoTools

Allow build to use Maven 2 or Maven 3.0

Details

  • Type: Bug Bug
  • Status: Resolved Resolved
  • Priority: Major Major
  • Resolution: Fixed
  • Affects Version/s: 8.0-M1
  • Fix Version/s: None
  • Component/s: admin
  • Labels:
    None

Description

Trying out maven 3.0.3 (for reported speed benifits). The build fails on our JAR files collector...

This is the subject of a proposal here:

  1. Text File
    GEOT-3615.-Dall.patch
    28/Jun/11 9:59 PM
    1 kB
    Ben Caradoc-Davies
  2. Text File
    patch_GEOT-3615.diff.patch
    16/Jun/11 1:46 AM
    0.8 kB
    Cliff Broadbent

Issue Links

is related to

Improvement - An improvement or enhancement to an existing feature or task. GEOS-4652 Support building GeoServer with Maven 3

  • Major - Major loss of function.
  • Closed - The issue is considered finished, the resolution is correct. Issues which are not closed can be reopened.

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Jody Garnett added a comment -
Could you attach your work on maven 3 to this issue; I may be able to scare up someone to look at this over the course of the week.
Show
Jody Garnett added a comment - Could you attach your work on maven 3 to this issue; I may be able to scare up someone to look at this over the course of the week.
Hide
Permalink
Andrea Aime added a comment -
The only patch I have is this one, apparently it was enough to let it build:

{code}

 git diff HEAD^..HEAD
diff --git a/pom.xml b/pom.xml
index 4a5bd96..da438e3 100644
--- a/pom.xml
+++ b/pom.xml
@@ -71,6 +71,7 @@
     <interactive.image>false</interactive.image>
     <online.skip.pattern>**/*OnlineTest.java</online.skip.pattern>
     <stress.skip.pattern>**/*StressTest.java</stress.skip.pattern>
+ <test.exclude.pattern>disabled</test.exclude.pattern>
     <test.maxHeapSize>512M</test.maxHeapSize>
     <src.output>${basedir}/target</src.output>
     <imageio.ext.version>1.1-RC1</imageio.ext.version>
{code}

The error message given about that missing property was really cryptic.
I don't remember if I tried a build from a clean local repo, nor I had time to double check the patch now.
Show
Andrea Aime added a comment - The only patch I have is this one, apparently it was enough to let it build: {code}  git diff HEAD^..HEAD diff --git a/pom.xml b/pom.xml index 4a5bd96..da438e3 100644 --- a/pom.xml +++ b/pom.xml @@ -71,6 +71,7 @@      <interactive.image>false</interactive.image>      <online.skip.pattern>**/*OnlineTest.java</online.skip.pattern>      <stress.skip.pattern>**/*StressTest.java</stress.skip.pattern> + <test.exclude.pattern>disabled</test.exclude.pattern>      <test.maxHeapSize>512M</test.maxHeapSize>      <src.output>${basedir}/target</src.output>      <imageio.ext.version>1.1-RC1</imageio.ext.version> {code} The error message given about that missing property was really cryptic. I don't remember if I tried a build from a clean local repo, nor I had time to double check the patch now.
Hide
Permalink
Cliff Broadbent added a comment -
Am looking into this one at the moment.

From a quick look, I'm thinking surefire doesn't like something, or it doesn't gracefully handle exceptions that are zero length strings.

If I work it out I'll post a patch.
Show
Cliff Broadbent added a comment - Am looking into this one at the moment. From a quick look, I'm thinking surefire doesn't like something, or it doesn't gracefully handle exceptions that are zero length strings. If I work it out I'll post a patch.
Hide
Permalink
Cliff Broadbent added a comment -
First attempt at a patch to convince Maven3 to play nicely with Geotools
Show
Cliff Broadbent added a comment - First attempt at a patch to convince Maven3 to play nicely with Geotools
Hide
Permalink
Cliff Broadbent added a comment -
And it helps when you attach the correct patch file...
Show
Cliff Broadbent added a comment - And it helps when you attach the correct patch file...
Hide
Permalink
Andrea Aime added a comment -
Cliff, how is your patch different from the one I added in the first comment of this jira?
Show
Andrea Aime added a comment - Cliff, how is your patch different from the one I added in the first comment of this jira?
Hide
Permalink
Cliff Broadbent added a comment -
The two patches achieve the same thing. Mine simply includes comments and comes in file form as I was unable to apply the code pasted in the initial comment.

It would appear the maven-surefire-plugin doesn't handle definition of a null value exclusion gracefully under maven 3.

I am raising a change proposal under the GeoTools wiki to move to Maven 3 as the transition appears reasonably straight forward.
Show
Cliff Broadbent added a comment - The two patches achieve the same thing. Mine simply includes comments and comes in file form as I was unable to apply the code pasted in the initial comment. It would appear the maven-surefire-plugin doesn't handle definition of a null value exclusion gracefully under maven 3. I am raising a change proposal under the GeoTools wiki to move to Maven 3 as the transition appears reasonably straight forward.
Hide
Permalink
Jody Garnett added a comment -
The fun part of the proposal will be updating the developers guide; especially the release procedure (if anything has changed in the deploy targets).
Show
Jody Garnett added a comment - The fun part of the proposal will be updating the developers guide; especially the release procedure (if anything has changed in the deploy targets).
Hide
Permalink
Ben Caradoc-Davies added a comment -
Cliff, your patch does not apply with "git apply". With patch it require fuzz 2. It also introduces a spurious unicode FEFF sequence before the XML declaration (look in a hex editor, use "od" or vi). And why use this filename? I think Andrea's is self-documenting.

And building with -Dall fails in gt-mark-wkt. Investigating.
Show
Ben Caradoc-Davies added a comment - Cliff, your patch does not apply with "git apply". With patch it require fuzz 2. It also introduces a spurious unicode FEFF sequence before the XML declaration (look in a hex editor, use "od" or vi). And why use this filename? I think Andrea's is self-documenting. And building with -Dall fails in gt-mark-wkt. Investigating.
Hide
Permalink
Ben Caradoc-Davies added a comment -
Got it: gt-mark-wkt has a parent mismatch: it is a submodule of unsupported, but declares its parent as library.
Show
Ben Caradoc-Davies added a comment - Got it: gt-mark-wkt has a parent mismatch: it is a submodule of unsupported, but declares its parent as library.
Hide
Permalink
Ben Caradoc-Davies added a comment -
Attached patch supersedes the earlier patch. It uses Andrea's version of the top-level pom fix and also patches gt-mark-wkt to support building with -Dall.
Show
Ben Caradoc-Davies added a comment - Attached patch supersedes the earlier patch. It uses Andrea's version of the top-level pom fix and also patches gt-mark-wkt to support building with -Dall.
Hide
Permalink
Ben Caradoc-Davies added a comment -
Cliff, as an aside, the U+FEFF at the start of your original patch is a UTF-16 byte order mark:
http://en.wikipedia.org/wiki/Byte_order_mark

Byte order marks are not recommended for UTF-8 documents. Furthermore, this is the wrong mark for this document, which is declared at UTF-8.

Looks like Notepad brain-damage to me.
Show
Ben Caradoc-Davies added a comment - Cliff, as an aside, the U+FEFF at the start of your original patch is a UTF-16 byte order mark: http://en.wikipedia.org/wiki/Byte_order_mark Byte order marks are not recommended for UTF-8 documents. Furthermore, this is the wrong mark for this document, which is declared at UTF-8. Looks like Notepad brain-damage to me.
Hide
Permalink
Andrea Aime added a comment -
That U+FEFF is added by a number of broken Windows programs, including Notepad, to mark the byte order. That breaks many xml readers as a result. See also: http://en.wikipedia.org/wiki/Byte_order_mark
Back when I was using windows I banned notepad and replaced it with Notepad2, which does not introduce these silly mistakes, does not force windows line endings and so on.
Show
Andrea Aime added a comment - That U+FEFF is added by a number of broken Windows programs, including Notepad, to mark the byte order. That breaks many xml readers as a result. See also: http://en.wikipedia.org/wiki/Byte_order_mark Back when I was using windows I banned notepad and replaced it with Notepad2, which does not introduce these silly mistakes, does not force windows line endings and so on.
Hide
Permalink
Jody Garnett added a comment -
Trying out the patch locally seems good so far.
Show
Jody Garnett added a comment - Trying out the patch locally seems good so far.
Hide
Permalink
Jody Garnett added a comment -
Commit applied to trunk as of r37549

(not sure if we want this applied to 2.7.x?)
Show
Jody Garnett added a comment - Commit applied to trunk as of r37549 (not sure if we want this applied to 2.7.x?)
Hide
Permalink
Cliff Broadbent added a comment -
Bizarre. I created the patch file using "git diff > patch.file" using git 1.7.6 from http://code.google.com/p/msysgit/downloads/list?can=3

Must have missed a switch somewhere.
Show
Cliff Broadbent added a comment - Bizarre. I created the patch file using "git diff > patch.file" using git 1.7.6 from http://code.google.com/p/msysgit/downloads/list?can=3 Must have missed a switch somewhere.

People

  • Assignee:
    Jody Garnett
    Reporter:
    Jody Garnett
Vote (0)
Watch (1)

Dates

  • Created:
    Updated:
    Resolved: