Filter by reset
- Everywhere
88 - Issues
23 - Pull Requests
19 - Comments
46
fix: update GitHub issue search to patched PHP 8.3.36
Summary Upgrade the demo from PHP 8.3.9 to patched PHP 8.3.36 (CVE-2026-109618 and CVE-2026-109619). The published manticoresearch/php-fpm:8.3.36 is based on Alpine 3.21.9 / musl, but this demo loads a prebuilt RHEL9 / glibc php-ext-model extension. A …
fix: update GitHub issue search to patched PHP 8.3.36
Change internal architecture of storing data
We should migrate to a system where we store each organization in a separate table. This approach will allow us to maintain the logic for searching one, two, or multiple repositories within a single organization. It will also help us narrow down query …
fix: update GitHub issue search to patched PHP 8.3.36
Show comments on state filter change for those issue state only (JOIN in the demo)
When we work with filters and search, we can currently filter by the state that only issues and pull requests have. However, when we change the state to open or close, we still search for comments everywhere. As discussed before, it would be a good idea …
size_S
Migrate to per-repo storage scheme
Empty list when adding a repository to the github demo
We should investigate why the list is empty on production servers but works correctly in the local environment.
Prevent 100k+ repos to be added
Since we aim for this GitHub issue search to be a demo project, there may be issues with big repositories due to indexing that takes a long time and consumes a lot of memory. To be fair and make it best suited as a demo project, we should set limits and …
est::size_S
Experiment with distributed table
Try to use a distributed table with issue and comment tables in it and CALL AUTOCOMPLETE on it to validate whether it works as expected or not.
Implement auto disabling of filters that have no data when we change the query for the current search.
After a crash or some kill -9 on the daemon, we noticed an issue with the indexers. They enter a state where they are "forever indexing," which leads to interface problems and a progress bar that never disappears. We should investigate and fix this issue, …
Migrate to per-repo storage scheme
As we discussed here we should try to migrate to table-per-repo scheme: https://github.com/manticoresoftware/manticore-github-issue-search/issues/25#issuecomment-2249352562 To do so we should first implement the logic that will help us dirrectly query the …
Enable all full-text search operators
We should consider adding logic that will allow us to find similar issues/comments in the same way we do in our Image demo.
est::size_M
Enable all full-text search operators
Consider enabling bigram_index to improve autocomplete
…Here The autocomplete results could be better if they included native support . It's currently impossible likely because bigram_index is not enabled. If we do it, the results may be better. E.g.: mysql> drop table if exists t; create table t(f text) min_ …
size_S
Enable all full-text search operators
Enable all full-text search operators
Change internal architecture of storing data
Enable operators for keyword search and disable query sanitizer
Change internal architecture of storing data
Notifications are not beeing send
We should check the cause of the issue when we are not receiving email notifications about indexation being done anymore.
bugsize_S
Change internal architecture of storing data
Let's add vector search to the demo using one of the most popular models. Implement the tokenizers Rust extension. Pick the best starting model. Add logic to handle various models and the option to adjust them at the start.
Change internal architecture of storing data
Long query leads to "Something went wrong"
When we input a very long query and use Fuzzy matching, we are experiencing a "Something went wrong" error. We should investigate this issue. An example of this is: https://github.manticoresearch.com/manticoresoftware/manticoresearch?query=after+daemon+ …
Change internal architecture of storing data
Various fixes and improvements to the Demo
improve exceptions handling, especially the exception which is thrown when the token has exceeded its limit experiment more with the facets UI to make it less confusing (grey out filters that would result in zero results etc.) deploy a new version of the …
Implement the query suggestion logic
We should move forward with enhancing the query suggestions panel and make it respond to the up and down arrow keys just like Google does. When users press the up or down arrows, it should set a new phrase in the input field and also change the focus …
Implement the query suggestion logic
Combine comment and issue in query suggestions
Currently, we only use issue responses to generate query suggestions when a user is typing. We should expand this approach by also checking comments and using an RPB-like algorithm to merge the results, returning a combined array. Additionally, to support …