IMAP UIDBATCHES Extension

The UIDBATCHES Extension An IMAP server advertises support for the UIDBATCHES extension by including the UIDBATCHES capability in the CAPABILITY response / response code.

UIDBATCHES Command

Arguments:: Message count per batch.OPTIONAL batch range.
Responses:: REQUIRED untagged response: UIDBATCHES
Result:: OK uidbatches completedNO command exceeds limitsBAD command unknown or arguments invalid

The UIDBATCHES command requests UID ranges that partition the messages in the currently selected mailbox into equally sized batches. The server returns these ranges in descending UID order, with batch 1 containing the highest UIDs (most recent messages), batch 2 containing the next highest set of UIDs, and so on. For a mailbox with M messages, requesting batches of size N returns UID ranges corresponding to the following sequence number ranges (where sequence numbers are ordered from 1 to M, with M being the most recent message):

Example Usage The following example demonstrates how a client uses UIDBATCHES to partition a mailbox into manageable batches: The server's response provides four UID ranges:

215295:99696
99695:20351
20350:7830
7829:1

Each range contains up to 2,000 messages, except the last range, which contains the remaining 823 messages. As new messages cannot appear within these UID ranges, the number of messages in each range will not increase. However, it may decrease as messages are deleted. It is only appropriate to resend UIDBATCHES if one of the following conditions is met:

A different mailbox has been selected
More than N/2 messages have been expunged from the mailbox (where N is the batch size)
More than N/2 new messages have been received into the mailbox

To prevent server overload, the client MUST NOT resend UIDBATCHES otherwise. Computing message batches may be resource-intensive for servers. The client can keep track of the number of EXPUNGE or VANISHED messages and re-run UIDBATCHES if many messages are deleted. As new messages arrive into the mailbox, the client should add these to a new message batch (starting at UID 215296 in the above example). Once N/2 or more new messages have been added to the mailbox, the client MAY ask for updated batches by re-running the UIDBATCHES command. The server SHOULD reject UIDBATCHES commands with a NO response with the LIMIT response code if the client exceeds this limit. Servers MAY choose to track client requests and mailbox state changes to enforce these restrictions and prevent resource abuse.

Response Format The server MUST reply with a UIDBATCHES response, even if no ranges are returned (see ). The UIDBATCHES response MUST include the tag of the command it relates to (similar to an ESEARCH response defined in ). The UID ranges in the response MUST be ordered in descending sequence, from the highest to the lowest UIDs.

Batch Sizes To ensure efficient server operation and prevent abuse, this extension enforces constraints on batch sizes. The design balances server efficiency requirements with the primary use case of working effectively with the UIDONLY mode , without creating a mechanism that circumvents the sequence number restrictions of that mode.

Batch Size Summary The following table provides a quick reference for batch size constraints: Batch Size Constraints

Constraint	Client Request	Server Response
Minimum	500 messages	Aim for >= 90% of requested size
Maximum	No limit	Never exceed requested size
Exception	-	May be smaller during mailbox changes

Key principles:

Clients MUST request at least 500 messages per batch (see Sections and )
Servers MUST NOT return more messages than requested
Servers SHOULD return batches close to the requested size (>= 90% when possible)
Exact batch sizes may vary due to implementation efficiency or mailbox changes

Minimum Batch Size The server MUST support batch sizes of 500 messages or larger. This minimum size prevents clients from misusing the extension to effectively reconstruct sequence numbers while still allowing reasonable batch operations. Note that clients MUST be prepared to handle batches smaller than requested, as detailed in . The server MUST respond with NO and a response code TOOFEW if the client uses a batch size smaller than the minimum allowed by the server:

Server Response Flexibility While servers SHOULD return batches that correspond exactly to the requested size, they have flexibility in specific circumstances to enable efficient implementations.

Hard Constraints The server MUST NOT return ranges that contain more than the number of messages per batch requested by the client. This is a strict upper bound that cannot be exceeded. If the requested batch size equals or exceeds the total number of messages in the mailbox, the server MUST return a single UID range spanning all messages.

Permitted Variations Servers MAY return fewer messages per range in two specific circumstances:

When doing so makes the implementation substantially simpler and/or more efficient
When there are changes in mailbox state during the execution of the UIDBATCHES command, particularly when messages are expunged

However, servers SHOULD NOT return batches that are substantially smaller than requested and SHOULD aim to stay within 90% of the requested size. This guideline reflects the fact that clients typically choose batch sizes based on their intended use, such as displaying a specific number of messages to users.

Dynamic Changes Mailbox state changes during UIDBATCHES execution can result in servers returning substantially fewer messages in each batch, particularly when message expungement reduces the overall mailbox size. Clients can detect these situations through the EXPUNGE, VANISHED, or EXISTS responses they receive.

Practical Implications Due to these flexibility provisions, servers may return batches of varying sizes. For instance, when returning 3 batches of a requested size of 1,000, one might contain 990 messages, another 977, and the third 1,000 messages. Clients MUST be prepared to handle such variations. When the total number of messages is not evenly divisible by the requested batch size, the final batch will contain the remainder. Therefore, the last batch in the mailbox (containing the lowest UIDs) will typically have fewer messages than requested.

Design Rationale These restrictions provide servers with implementation flexibility while preventing clients from misusing the extension to effectively reconstruct sequence numbers. also outlines some reasoning for these limitations. The flexibility regarding batch sizes is designed to enable efficient server implementations while maintaining predictable behavior for clients. This leeway is not intended as a general permission to return arbitrarily sized batches but rather to accommodate implementation constraints and dynamic mailbox changes.

UIDs The server MAY return UID ranges with UIDs that do not exist on the server. As a result, the client MUST NOT make assumptions about the existence of messages. If the server returns the response: there may not be any messages on the server with the UIDs such as 163886, 99703, 99696, etc. The range 163886:99703 will span approximately the requested number of messages (may be less, see ), but its start and end UIDs may not correspond to messages on the server. This gives the server implementation some flexibility as to which UID ranges to return. They might, e.g., return 163886:99697 and 99696:20358 instead of 163886:99703 and 99696:20358 -- assuming that there are no messages in the range 99702:99697. If there are fewer messages in the mailbox than the requested batch size, the server would return a single batch that contains all messages in the mailbox. When applying the flexibility described above to the last batch in the mailbox, ending that batch with UID 1 makes it unambiguous to the client that this range is in fact the last range. For example, if the message with the lowest UID is 302, the server can return 7829:1 instead of 7829:302.

Batch Ranges A client can optionally provide a batch range. The server limits its response to UID ranges corresponding to the specified batch indices. For example, if the client sends: for a mailbox with 100,000 messages, the server would return the 10th to 20th batches. The 10th batch would correspond to message sequence numbers `82000:80001` and the 20th batch would correspond to message sequence numbers `62000:60001`. Batches start at the highest UIDs, i.e., batch 1 is the batch with the highest UIDs. The UID ranges that the server returns would still split the mailbox's messages into batches of the requested size (2,000 in the example). If the client requests more batches than exist on the server, the server would return those that do exist. For example, if the client sends: and the selected mailbox has 7,000 messages, the server would then return a UIDBATCHES response with only 4 UID ranges. Batch ranges such as 1:4 in the above example MUST be ordered lowest to highest, i.e., be sent as 1:4 and not as 4:1. Servers MUST reject batch ranges that are in the wrong order with BAD and a response code CLIENTBUG: If the client requests a range of batches that do not exist on the server, the server MUST still return an empty response. See . The number of messages per batch returned by the server may be approximate, as detailed in . As a result, if the client needs to request consecutive batch ranges such as 1:100, 101:200, 201:300, and so on, the client may want to make these batch ranges overlap by, e.g., requesting 1:100, 100:200, and 200:300. While the UIDs returned may not correspond to existing messages (as described in ) and mailbox state can change between requests, checking whether the returned UID ranges overlap can help clients detect potential inconsistencies, though how to handle such situations depends on the specific client implementation requirements. Clients MUST NOT request batch ranges that span more than 100,000 messages, i.e., the number of batches multiplied by the batch size MUST NOT be larger than 100,000. This restriction applies only when a batch range is specified; when no batch range is provided, the client is requesting all batches but the server may limit its response as described in . The server MAY reject UIDBATCHES commands with a NO response with the TOOMANY response code if the client exceeds this limit.

Empty Responses When the client issues any valid UIDBATCHES command and the mailbox is empty, the server MUST reply with a UIDBATCHES response, for example: If the client requests a range of batches that do not exist, the server MUST reply with an empty UIDBATCHES response. If the mailbox has 7,000 messages, and the client sends: the server would respond with:

Large Mailboxes The server may not be able to return all UID ranges if the mailbox contains an extremely large number of messages. The server MUST at least support returning UID ranges spanning 100,000 messages. See for details on this limit. If the server cannot return all of the requested UID ranges, it MUST respond with a NO response with the TOOMANY response code. Notably, when the client requests all UID ranges and the mailbox has more than 100,000 messages, the server MAY reply with a NO response. For example: The client should know what the message count in the mailbox is, and if the message count exceeds 100,000, it may choose to always request batch ranges as discussed in instead of requesting all batches.

Interaction with MESSAGELIMIT Extension When the server supports both the MESSAGELIMIT and UIDBATCHES extension, the client SHOULD request batches no larger than the specified maximum number of messages that can be processed in a single command. The client MAY choose to use a smaller batch size. Additionally, since servers MAY limit the number of UIDs returned in response to UIDBATCHES, it is reasonable to assume that they would at most return N UIDs, where N is the limit the server announced as its MESSAGELIMIT.

Interaction with UIDONLY Extension The UIDBATCHES extension allows clients to create UID ranges for message batches even when the connection operates in UIDONLY mode, which otherwise doesn't allow for using message sequence numbers. This interaction is particularly important because the UIDONLY extension disallows the use of sequence numbers. While the PARTIAL extension provides paged SEARCH and FETCH operations, some clients need to predetermine UID ranges for batches upfront. UIDBATCHES enables such clients to use the same overall batching strategy, regardless of whether the server supports UIDONLY, PARTIAL, or neither, making client implementations simpler and more consistent. When operating in UIDONLY mode, clients SHOULD use UIDBATCHES to determine appropriate UID ranges for batch operations rather than attempting to construct batches using sequence-number-based approaches that would violate UIDONLY restrictions. The batch size constraints defined in serve dual purposes: ensuring server efficiency and preventing UIDBATCHES from becoming a mechanism to effectively reconstruct sequence numbers. Without these constraints, clients could request very small batch sizes to obtain fine-grained positional information about messages, which would circumvent the sequence number restrictions of UIDONLY mode. The extension is designed specifically to support the legitimate need of predetermining message batches upfront while maintaining the architectural intent of UIDONLY mode.

Interaction with SEARCHRES Extension UIDBATCHES is not a SEARCH nor UID SEARCH command. Servers that support SEARCHRES MUST NOT store the result of UIDBATCHES in the $ variable.