Bibliographic record
Abstract
Ad Hoc Transactions: What They Are and Why We Should CareMost database research papers are prescriptive.They identify a technical problem and show us how to solve it.They present new algorithms, theorems, and evaluations of prototypes.Other papers follow a different path; they are descriptive rather than prescriptive.They tell us how data systems behave in practice and how they are actually used.They employ a different set of tools, such as surveys, software analyses, or user studies.These papers are much rarer at database research conferences, and they're all the more valuable for that.The paper I'm introducing here, Many Faces of Ad Hoc Transactions, by Tang et al., is of this descriptive type.It presents an analysis of open source database applications, focusing on how these applications synchronize concurrent operations.According to the authors, this analysis is the result of five person-years of effort.(I suspect this is one of the reasons work like this is not more common.)This paper originally appeared at SIGMOD 2022, 2 and it extends a (short) research thread that first came to my attention through earlier work by Warszawski, Bailis, and others. 1,3 "Wait", I hear you say."Database systems provide transactions, and applications use transactions to synchronize their database accesses.What's to study here?"If this is your immediate reaction, I urge you to read this paper.Transactions are, of course, widely used, but database applications also have other tools at their disposal for coordinating concurrent actions.These include synchronization primitives provided by database systems, such as explicit locks, as well as synchronization mechanisms in the application itself.Tang et al. refer to applications' use of these mechanisms as "ad hoc transactions" to distinguish them from transactions implemented in the database system.As it turns out, ad hoc transactions are quite common.Tang et al. analyzed eight popular database applications and found ad hoc transactions in all of them-more than 90 examples of ad hoc transactions in all.This ad hoc transaction corpus is the focus of the paper.Let's consider one example from the corpus, as an illustration.It is an ad hoc transaction used in the Discourse forum application to update the content of a forum post.It spans two application-level HTTP requests.In the first, a client requests the current content of a forum post for editing.In response, the server-side application uses a read-only database transaction to retrieve the post content and version number and returns them to the client.The client edits the content and then issues a second HTTP request to install the new post content, providing the original version number.In response, the server-side application first uses a read-only database transaction to verify that the version number has not changed since the content was originally retrieved.If it has, the update fails.Otherwise, the application uses another database transaction to install the new content and update the version number.The application also uses an explicit lock on the post identifier to prevent concurrent updates to the post between the version check transaction and the update transaction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.016 | 0.003 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".