The Sage Trac Server#
Sage development is scheduled to move to GitHub in February 2023. The exact date will be announced in https://groups.google.com/g/sage-devel. After the transition, some parts of this guide (especially those related with the Sage Trac server) will become obsolete and be updated according to the new workflow on GitHub. See our transition guide from Trac to GitHub for the preliminary version of the workflow.
Sometimes you will only want to work on local changes to Sage, for your own private needs. However, typically it is beneficial to share code and ideas with others; the manner in which the Sage project does this (as well as fixing bugs and upgrading components) is in a very collaborative and public setting on the Sage Trac server (the Sage bug and enhancement tracker).
The purpose of the Sage trac server is to
Provide a place for discussion on issues and store a permanent record.
Provide a repository of source code and all proposed changes.
Link these two together.
There is also a wiki for more general organizational web pages, like Sage development workshops.
Thus if you find a bug in Sage, if you have new code to submit, want to review new code already written but not yet included in Sage, or if you have corrections for the documentation, you should post on the trac server. Items on the server are called tickets, and anyone may search or browse the tickets. For a list of recent changes, just visit the Sage trac timeline.
Sage development is scheduled to move to GitHub in February 2023. All functions of our Trac server will be taken over by our main repository, https://github.com/sagemath/sage.
Obtaining an Account#
If you do not have an account on GitHub yet, choose a user name and create an account.
Using your GitHub account, you can log in to:
the Sage trac server, so that you can open tickets and participate in discussions on existing tickets.
On the Sage trac server, click the link “GitHub Login” in the top right corner and follow the prompts.
Within the Sage trac server, your GitHub user name will be prefixed with the letters
gh-(which stand for “GitHub”).
GitLab, where the mirror repository sagemath/sage accepts Merge Requests.
In GitLab, click the button “Sign in / Register” in the top right corner, and then use the button “Sign in with GitHub” and follow the prompts.
Users with legacy sage-trac accounts (account names not starting with “gh-”) should use the Login link. Do not to use GitHub login, as it will be treated as a separate user from their original account (unless you actively prefer to switch).
If you think you have found a bug in Sage, here is the procedure:
Search through our Google groups for postings related to your possible bug (it may have been fixed/reported already):
Similarly, you can search The Sage Trac Server to see if anyone else has opened a ticket about your bug.
If you do not find anything, and you are not sure that you have found a bug, ask about it on sage-devel. A bug report should contain:
An explicit and reproducible example illustrating your bug (and/or the steps required to reproduce the buggy behavior).
The version of Sage you run, as well as the version of the optional packages that may be involved in the bug.
Describe your operating system as accurately as you can and your architecture (32-bit, 64-bit, …).
You might be asked to open a new ticket. In this case, follow the Guidelines for Opening Tickets.
Thank you in advance for reporting bugs to improve Sage in the future!
Guidelines for Opening Tickets#
In addition to bug reports (see Reporting Bugs), you should also open a
ticket if you have some new code that makes Sage a better tool. If you have a
feature request, start a discussion on
sage-devel first, and then if there
seems to be general agreement that you have a good idea, open a ticket
describing the idea.
Do you already have a trac account? If not, click here.
Before opening a new ticket, consider the following points:
Make sure that nobody else has opened a ticket about the same or closely related issue.
It is much better to open several specific tickets than one that is very broad. Indeed, a single ticket which deals with lots of different issues can be quite problematic, and should be avoided.
Be precise: If foo does not work on OS X but is fine on Linux, mention that in the title. Use the keyword option so that searches will pick up the issue.
The problem described in the ticket must be solvable. For example, it would be silly to open a ticket whose purpose was “Make Sage the best mathematical software in the world”. There is no metric to measure this properly and it is highly subjective.
For bug reports: the ticket’s description should contain the information described at Reporting Bugs.
If appropriate, provide URLs to background information or sage-devel conversation relevant to the problem you are reporting.
When creating the ticket, you may find useful to read The Ticket Fields.
Unless you know what you are doing, leave the milestone field to its default value.
The Ticket Fields#
When you open a new ticket or change an existing ticket, you will find a variety of fields that can be changed. Here is a comprehensive overview (for the ‘status’ field, see The Status of a Ticket):
Reported by: The trac account name of whoever created the ticket. Cannot be changed.
Owned by: Trac account name of owner, by default the person in charge of the Component (see below). Generally not used in the Sage trac.
Type: One of
enhancement(e.g. a new feature),
defect(e.g. a bug fix), or
Priority: The priority of the ticket. Keep in mind that the “blocker” label should be used very sparingly.
Milestone: Milestones are usually goals to be met while working toward a release. In Sage’s trac, we use milestones instead of releases. Each ticket must have a milestone assigned. If you are unsure, assign it to the current milestone.
Component: A list of components of Sage, pick one that most closely matches the ticket.
Keywords: List of keywords. Fill in any keywords that you think will make your ticket easier to find. Tickets that have been worked on at Sage days
NN(some number) ofter have
Cc: List of trac user names to Cc (send emails for changes on the ticket). Note that users that enter a comment are automatically substcribed to email updates and don’t need to be listed under Cc.
Merged in: The Sage release where the ticket was merged in. Only changed by the release manager.
Authors: Real name of the ticket author(s). Set this field only if you intend to provide code.
Reviewers: Real name of the ticket reviewer(s).
Report Upstream: If the ticket is a bug in an upstream component of Sage, this field is used to summarize the communication with the upstream developers.
Work issues: Issues that need to be resolved before the ticket can leave the “needs work” status.
Branch: The Git branch containing the ticket’s code (see Branching Out). It is displayed in green color, unless there is a conflict between the branch and the latest beta release (red color). In this case, the branch should be merged or rebased.
Dependencies: Does the ticket depend on another ticket? Sometimes, a ticket requires that another ticket be applied first. If this is the case, put the dependencies as a comma-separated list (
#1234, #5678) into the “Dependencies:” field.
Stopgaps: See Stopgaps.
The Status of a Ticket#
The status of a ticket appears right next to its number, at the top-left corner of its page. It indicates who has to work on it.
new – the ticket has only been created (or the author forgot to change the status to something else).
If you intend to work on the code yourself, put your name in the Authors field, or leave a comment to say so. It could avoid having two persons doing the same job.
needs_review – the code is ready to be peer-reviewed. If the code is not yours, then you can review it. See The reviewer’s check list.
needs_work – something needs to be changed in the code. The reason should appear in the comments.
needs_info – somebody has to answer a question before anything else can happen. It should be clear from the comments.
positive_review – the ticket has been reviewed, and the release manager will close it.
The status of a ticket can be changed using a form at the bottom of the ticket’s page. Leave a comment explaining your reasons whenever you change it.
When Sage returns wrong results, two tickets should be opened:
A main ticket with all available details.
A “stopgap” ticket (e.g. trac ticket #12699)
This second ticket does not fix the problem but adds a warning that will be printed whenever anyone uses the relevant code. This, until the problem is finally fixed.
To produce the warning message, use code like the following:
from sage.misc.stopgap import stopgap stopgap("This code contains bugs and may be mathematically unreliable.", TICKET_NUM)
TICKET_NUM by the ticket number for the main ticket. On the main
trac ticket, enter the ticket number for the stopgap ticket in the “Stopgaps”
field (see The Ticket Fields). Stopgap tickets should be marked as
If mathematically valid code causes Sage to raise an error or crash, for example, there is no need for a stopgap. Rather, stopgaps are to warn users that they may be using buggy code; if Sage crashes, this is not an issue.
Working on Tickets#
If you manage to fix a bug or enhance Sage you are our hero. See Sage Development Process for making changes to the Sage source code, uploading them to the Sage trac server, and finally putting your new branch on the trac ticket.
After pushing a branch to a ticket, the ticket will show badges linking to results of automated tests that run on the patchbot and other tests that run on GitHub Actions.
A linting workflow runs on all pushes to a branch on Trac. It checks that the code of the current branch adheres to the style guidelines using Pycodestyle (in the
pycodestyle-minimalconfiguration) and Relint.
In order to see details when it fails, you can click on the badge and then select the most recent workflow run.
The incremental build and test workflow on GitHub Actions builds Sage for the current branch (incrementally on top of an installation of the
developbranch) and runs the test. Note that in contrast to the patchbot, the ticket branch is not merged into the current beta version.
Details are again available by clicking on the badge.
The automatic workflow runs on a container based on
ubuntu-focal-standard. To request a run of the workflow on a different platform, you can issue a workflow_dispatch. You can select any of the platforms for which a prebuilt container image exists.
The build documentation workflow on GitHub Actions builds the HTML documentation for the current branch.
If you click on the badge, you get the HTML output of the successful run. The idea is to use this to easily inspect changes to the documentation without the need to locally rebuild the docs yourself. If the doc build fails, you can go to the Actions tab of sagemath/sagetrac-mirror repo and choose the particular branch to see what went wrong.
The patch buildbot will automatically test your ticket. See trac ticket #wiki/patchbot for more information about its features and limitations. Make sure that you look at the log, especially if the patch buildbot did not give you the green blob.
Sage development is scheduled to move to GitHub in February 2023. After the move, the patch buildbot will no longer be available; the three workflows above are considered a full replacement. If you miss any features of the patch buildbot, please report this in trac ticket #33457.
The following are some other relevant issues:
Every bug fixed should result in a doctest.
This is not an issue with defects, but there are many enhancements possible for Sage and too few developers to implement all the good ideas. The trac server is useful for keeping ideas in a central place because in the Google groups they tend to get lost once they drop off the first page.
If you are a developer, be nice and try to solve a stale/old ticket every once in a while.
Some people regularly do triage. In this context, this means that we look at new bugs and classify them according to our perceived priority. It is very likely that different people will see priorities of bugs very differently from us, so please let us know if you see a problem with specific tickets.
Reviewing and Closing Tickets#
Tickets can be closed when they have positive review or for other reasons. To learn how to review, please see The reviewer’s check list.
Only the Sage release manager will close tickets. Most likely, this is not you nor will your trac account have the necessary permissions. If you feel strongly that a ticket should be closed or deleted, then change the status of the ticket to needs review and change the milestone to sage-duplicate/invalid/wontfix. You should also comment on the ticket, explaining why it should be closed. If another developer agrees, he sets the ticket to positive review.
A related issue is re-opening tickets. You should refrain from re-opening a ticket that is already closed. Instead, open a new ticket and provide a link in the description to the old ticket.
Reasons to Invalidate Tickets#
One Issue Per Ticket: A ticket must cover only one issue and should not be a laundry list of unrelated issues. If a ticket covers more than one issue, we cannot close it and while some of the patches have been applied to a given release, the ticket would remain in limbo.
No Patch Bombs: Code that goes into Sage is peer-reviewed. If you show up with an 80,000 lines of code bundle that completely rips out a subsystem and replaces it with something else, you can imagine that the review process will be a little tedious. These huge patch bombs are problematic for several reasons and we prefer small, gradual changes that are easy to review and apply. This is not always possible (e.g. coercion rewrite), but it is still highly recommended that you avoid this style of development unless there is no way around it.
Sage Specific: Sage’s philosophy is that we ship everything (or close to it) in one source tarball to make debugging possible. You can imagine the combinatorial explosion we would have to deal with if you replaced only ten components of Sage with external packages. Once you start replacing some of the more essential components of Sage that are commonly packaged (e.g. Pari, GAP, lisp, gmp), it is no longer a problem that belongs in our tracker. If your distribution’s Pari package is buggy for example, file a bug report with them. We are usually willing and able to solve the problem, but there are no guarantees that we will help you out. Looking at the open number of tickets that are Sage specific, you hopefully will understand why.
No Support Discussions: The trac installation is not meant to
be a system to track down problems when using Sage. Tickets should
be clearly a bug and not “I tried to do X and I couldn’t get it to
work. How do I do this?” That is usually not a bug in Sage and it
is likely that
sage-support can answer that question for you. If
it turns out that you did hit a bug, somebody will open a concise
and to-the-point ticket.
Solution Must Be Achievable: Tickets must be achievable. Many times, tickets that fall into this category usually ran afoul to some of the other rules listed above. An example would be to “Make Sage the best CAS in the world”. There is no metric to measure this properly and it is highly subjective.
The Release Process#
The Sage Release Manager uses the following procedure to make releases, as of 2022.
Beta Release Stage: For preparing a new beta release or the first release
candidate, all positively reviewed tickets with the forthcoming release
milestone are considered. Tickets that have unmerged dependencies are ignored.
The Release Manager merges tickets in batches of 10 to 20 tickets, taking the
ticket priority into account. If a merge conflict of a ticket to the Release
Manager’s branch occurs, the ticket is set back to “needs work” status by the
Release Manager, and the list of the tickets already merged to the Release
Manager’s branch is posted. The author of the ticket needs to identify
conflicting tickets in the list, make merge commits and declare them as
dependencies, before setting back to “positive review” status. Each batch of
merged tickets then undergoes integration testing. If problems are detected, a
ticket will be set back to “needs work” status and unmerged. When a batch of
tickets is ready, the Release Manager closes these tickets and proceeds to the
next batch. After a few batches, a new beta release is tagged, pushed to the
develop branch on the main git repository, and announced on
Release Candidate Stage: After the first release candidate has been made, the project is in the release candidate stage, and a modified procedure is used. Now only tickets with a priority set to “blocker” are considered. Tickets with all other priorities, including “critical”, are ignored. Hence if a ticket is important enough to merit inclusion in this stage, it should be set to “blocker”.
Blocker Tickets: The goal of the release process is to make a stable release of high quality. Be aware that there is a risk/benefit trade-off in merging a ticket. The benefit of merging a ticket is the improvement that the ticket brings, such as fixing a bug. However, any code change has a risk of introducing unforeseen new problems and thus delaying the release: If a new issue triggers another release candidate, it delays the release by 1-2 weeks. Hence developers should use “blocker” priority sparingly and should indicate the rationale on the ticket. Though there is no one fixed rule or authority that determines what is appropriate for “blocker” status,
Tickets introducing new features are usually not blockers – unless perhaps they round out a set of features that were the focus of development of this release cycle.
Tickets that make big changes to the code, for example refactoring tickets, are usually not blockers.
Final Release: If there is no blocker ticket for the last release
candidate, the Release Manager turns it to the final release. It is tagged with
the release milestone, and announced on