Difference between revisions of "Development Resources/Automatic IP Log"
(→Problems with the IP Log?)
|Line 26:||Line 26:|
=Using the Automated IP Log=
=Using the Automated IP Log=
==Correcting Errors In General==
==Correcting Errors In General==
Revision as of 22:26, 20 September 2010
The Foundation automatically produces IP logs for each project (bug 220977). This page describes how the automatic IP Log is generated (by the system) and maintained (by the project team) and processed (by Eclipse Legal).
- 1 What is an IP Log?
- 2 What is the Automated IP Log Tool?
- 3 Using the Automated IP Log
- 3.1 Correcting Errors In General
- 3.2 Third-Party Code
- 3.3 Unused Third-Party Code
- 3.4 Works With and Pre-Req Dependencies
- 3.5 Pending Contribution Questionnaires
- 3.6 Committers
- 3.7 Contributors
- 3.8 Keeping the IP Log Current
- 3.9 Choosing the Set of Projects
- 3.10 Submitting the IP Log to Eclipse Legal
- 3.11 Problems with the IP Log?
What is an IP Log?
An IP Log is a record of the IP contributions to a project. This includes such as a list of all committers, past and present, that have worked on the code and (especially) those who have made contributions to the current code base. The expected contents of a log is described on the Project IP Log page.
What is the Automated IP Log Tool?
The Automated IP Log Tool automatically generates an IP Log using information that is available to the Eclipse Foundation. The list of committers, for example is generated using information provided by the Dash project which itself pulls information out of eclipse.org CVS and SVN repositories.
To fully leverage the value of the Automated IP Log Tool, you need to:
- Keep your project metadata up-to-date
- Mark IP Contributions in Bugzilla
- Create contribution questionnaires (CQs) where appropriate
Your project's metadata is used to determine the identities of the source code repositories that Dash needs to scan to find out committer information. Specifically, you need to specify, in the "source repository" section, a list of paths to CVS and SVN repository locations. With this information specified, not only will the Automated IP Log tool be able to build the "Committers" section of the IP Log for you, but your project's Project Summary page will also be updated to accurately reflect your project's commit activity.
Alternatively, you can flag an entire Bugzilla entry with +iplog. Doing so, however, indicates to the Automated IP Log tool that every single comment made by a non-committer in the bug report represents a potential contribution. For your own sanity, it's a good practice to ask contributors to provide and attach patches that can be individually marked. Marking an entire bug represents an ongoing maintenance issue as new comments added to the bug from non-committers will show up in the generated log (bogus entries will have to be excluded - see below).
The Third-Party Software section of the log is populated from our IPZilla Bugzilla instance. The IP Team will mark your contributions in such a way that they will appear in the log. If third party software is not appearing properly, contact the EMO IP Team to make corrections.
The rest of this document is concerned with tuning the output of the Automated IP Log generator.
As of June 2009, we have started work on a new generation of tools that will be ready in the Fall of 2009. These tools will use the mechanisms indicated above.
Using the Automated IP Log
To start with the Automated IP Log, you need to first navigate to the page. The URL is of the form: http://www.eclipse.org/projects/ip_log.php?projectid=[projectId] (e.g. http://www.eclipse.org/projects/ip_log.php?projectid=dsdp.rtsc). A link to the IP Log is provided in the project summary page for each project.
Note that the log found at this URL is marked "Tentative". This is automatically-generated content that is only considered official when it is actually submitted. To submit the log for review, you must log-in (using the link at the bottom of the page), correct any errors that you find (see below), and click "Submit".
Note that you can adjust the set of projects that contribute data to the log by clicking the "Adjust" ("the set of projects to match the upcoming release") button. A log can be submitted for multiple projects; you might do this if your project provides a "roll-up" release and download that includes code from one or more subprojects.
Note that, while you can submit a log for multiple projects, you can only edit the log for a single project. If you are doing a roll-up release, first edit the logs for each of the individual projects, and then submit the log for the combination.
When the log is submitted, it is reviewed by the IP Team. They may have questions, or ask for clarifications. You may be asked to resubmit the log after you have made some changes. When the process is complete, the IP Team will email you a PDF that contains the official log; you should make this available for your consumers and adopters to review.
Correcting Errors In General
In order to correct errors in the IP Log, you must:
- Be a committer on the project
- Be looking at an IP Log for a single project (because if there are multiple projects, we can't figure out where to record the corrections)
- Have the same email address for your Foundation email and your bugzilla email.
- Be logged in to the Eclipse website. The login mechanism uses the Friends of Eclipse login, so when you click "login" in the bottom menu, you will be taken to the Friends of Eclipse login. You do not need to be a Friend to use this login page: use your bugzilla login.
When all of that is true, extra correction form fields will appear on the page along with a "Make These Corrections" submit button at the bottom. Like any html form, fill it out and press the button to save the corrections.
The Third-Party Code section of the IP Log shows the approved Contribution Questionnaires from ipzilla. The CQs listed in this table are those that:
- state=approved, or
- RESOLVED/FIXED or CLOSED/FIXED (regardless of state), and
- contain the keyword "nonepl"
- Missing CQ. If a CQ is missing from this list, (a) check its status, resolution, and state: make sure that they match one of the above cases, and (b) check to ensure "nonepl" keyword is included in order to identify the code as third party. If not, contact Eclipse Legal.
- Extra CQ. If a CQ is listed but does not belong to this project, then contact Eclipse Legal to have the ipzilla database corrected.
- Missing or Incorrect License. If a CQ is missing the license information, contact Eclipse Legal to have the correct license information added to the database.
Unused Third-Party Code
If the project has CQs that approve the use of certain third-party code, but that code is not currently being used for this release, it is listed in the separate Unused Third-Party Code table. The CQs listed in this table are those that have one of the following keywords:
- obsolete - this third-party code was used in previous releases, but it has been superseded or is obsolete and thus is no longer being used in this release and there is no plan to use it in a future release.
- unused - the project requested approval of this third-party code in anticipation of using this code, but at this time the project is not using this code in this release. The project reserves the right to use the code in a future release.
- withdrawn - the project requested approval of this third-code code but subsequent to receiving approval and prior to using the code, decided not to use said code. This code has not been used in any release, is not being used in this release, and there is no plan to use it in a future release.
- CQ Is Being Used. If a CQ is listed as unused but is actually being used, contact Eclipse Legal to have the keywords on CQ removed/changed.
- CQ Should Be Unused. If a CQ should be in the unused list instead of one of the other lists, contact Eclipse Legal to have the appropriate keyword added.
Works With and Pre-Req Dependencies
If the project has Works With, Pre-Req, or Exempt Pre-Req Dependencies, these will be listed in a separate table. These are also Contribution Questionnaires from ipzilla. These CQs are:
- Missing Pre-Req. If an pre-req is missing from this list, first check that to see if there is a CQ for it. If there is no CQ, submit one (using the portal). If there is a CQ, check that its state is "workswith", "prereq", or "exempt_prereq" - if not, contact Eclipse Legal.
Pending Contribution Questionnaires
If the project has any pending and not yet approved or rejected Contribution Questionnaires, these are listed in a separate table. These represent potential IP issues and thus must be considered carefully during a Release Review.
- Unneeded CQ. If the code represented by a pending CQ is no longer needed by the project, use ipzilla to withdraw the CQ.
The Committers section of the IP Log uses two information sources to generate the two lists:
- The list of all committers who have committed code to any of the source repositories from the Commits Explorer. This is the basis of the "Past and Present" list. The commits explorer list is driven by the project meta data
- The list of all past and present committers from the Foundation's internal database of committers and projects. Committers here but not in in the "Past and Present" list are put on the "Never Active" list.
- Should Be Active. If a committer is listed as "Never Active" but should be listed in the active "Past and Present" table, then you should check the project meta data via the portal - there are directories missing from the "source repository" item. If the project meta data is correct, you can use the "make active" checkbox to submit a correction.
- Question Mark. If a "xxx???" person is listed as a committer, that means that the unix login "xxx" committed code to the source code repository, but there is no record of "xxx" as having been a committer on the project. This is a potential IP cleanliness problem, so contact the EMO to resolve it (do not just use the "exclude" checkbox correction without understanding why the data is incorrect).
- Extra Committer. If someone is listed as a "Never Active" committer, but you know that they were never even a committer at all, then there is an incorrect record in the Foundation database - contact the EMO. On the other hand, if someone is listed as a "Past and Present" committer, but you know that they were never a committer, then something larger is awry because there are source code check-ins by that person. Use the commits explorer to determine what check-ins they made and contact the EMO if there are any errors. In both cases, after resolving the issue with the EMO, use the "exclude" checkbox to correct the IP log.
- Missing Committer. If someone who is/was a committer with a unix account and is missing from both the "Past and Present" and the "Never Active" tables, then the first thing to check is the "source repository" project meta data item via the portal - usually the problem is one or more unlisted source directories.
The Contributors section of the IP Log uses one information source: bugzilla (driven by the project meta data
- all people who were not committers when they attached:
- an attachment to a bug
- and the attachment is flagged with "iplog+"
- and is attached to a RESOLVED/FIXED, VERIFIED/FIXED, or CLOSED/FIXED bug
- all people who were not committers when they commented on a bug:
- that is flagged with "iplog+"
- and RESOLVED/FIXED, VERIFIED/FIXED, or CLOSED/FIXED
In plain English, this means there are two different ways to mark a contribution in bugzilla. If the contribution is an attachment to a bug such as a patch, the attachment itself should be marked with the iplog+ flag. If the contribution is in the form of a comment in the bug, then the iplog+ flag should be set on the bug itself. Since this will mark all non-committers who commented on the bug as contributors, you typically then need to use the exceptions mechanism to exclude comments that are not contributions. Typically these two mechanisms are mutually exclusive: if the contribution is an attachment, you do not also need to set iplog+ flag on the bug itself (unless the bug also has other contributions in the form of comments).
- No Name. If the contributor record for person X shows an email address but no name, then person X has not entered a name in bugzilla. Have X go to bugzilla Preferences and enter their correct name.
- Error Messages. If the tentative IP log shows errors (red error messages) in product name or component name, then the team should check their project meta data via the portal - the product name and (optional) component names must be an exact match with bugzilla, capitalization and all.
- Extra Contributor. If a person (person X) is listed as a Contributor but the team knows that their patch was never used/is not currently in use, there are three solutions: (1; preferred) remove the "iplog+" flag from the bugzilla attachment, or (2) use "exclude" correction to modify the IP log (if contributor status is based on a comment instead of an attachment).
- Missing Contributor. If person X should be listed as a Contributor but is not, then be sure that person X has attached a patch to a fixed bug and that patch is flagged with "iplog+". All external code contributions should come through bugzilla patches, so if a code contribution first comes through email, the contribution must also add a patch. Note that person X must add the patch because if person Y adds the patch for person X, then person Y will be listed as the contributor.
- Committer is Listed. If committer X is listed as a Contributor, check the following:
- First, a committer can be listed as a contributor if he or she was a contributor, contributed some code, and then later became a committer. He/she will be listed as a contributor for the patches he/she submitted before becoming a committer.
- Check that person X's email address in bugzilla matches any one of their email addresses on file with the Foundation. Have person X use the the portal to check to verify their email addresses. If their bugzilla address does not match the Foundation committer records, we have no way to associate their bugzilla records with their committer status.
Keeping the IP Log Current
Keeping the automatic IP log current is fairly simple:
- Make sure that all third-party code and dependencies are filed as CQs.
- If a non-committer patch is accepted, mark it with the iplog+ flag.
- If a non-committer bug without a patch is accepted, mark the whole bug with the iplog+ flag and then go to the automatic IP log to exclude the non-contribution comments.
- Use the portal to deactivate inactive committers
Choosing the Set of Projects
The simplest case is an IP Log for a single project. However, many project teams work together to create a unified release across a set of projects. In this case, the IP Log must also be an aggregated/unified IP Log across the same set of projects. At the bottom left of the IP Log page is a "adjust the set of projects" link. That link leads to the selector page. Use the checkboxes to select a set of projects and generate a unified IP Log for all of those projects.
Submitting the IP Log to Eclipse Legal
Single or unified-multi-project, either way the IP Log can be submitted to Eclipse Legal for review using the link at bottom center of the page. Click that, fill out the reason, and voila, the a frozen copy of the log is sent to Legal.
You must send your IP log to Eclipse Legal at least three weeks in advance of your scheduled Release Review to guarantee your review date. Sending your IP log in late will jeopardize your review date due to Eclipse Legal's workload.
Problems with the IP Log?
If you're having trouble updating the IP log or if the IP log is not showing enough information, please file a bug (use this link) and we'll work hard to make things right.
This page is moderated by the EMO