Right now we only get an exception, so we built a splitting mechanism by ourselfes, but it would be great if your lib would adjust the message size automatically. Also sometimes we are still hitting the limits (even though we are calculating all the bytes by ourselfes), so maybe you have a better solution?
This is our solution (we are rebuilding the message on the client side afterwards):
max_message_size = 190_000 # Limit is 256kb, more than 190_000 doesn't work sometimes.. dunno why
chars = message.to_json.to_s.chars
message_chunks = []
while chars.any?
chunks = []
chunksize = 0
while chars.any? && chunksize < max_message_size
if (chunksize + chars[0].bytesize) > max_message_size
break
else
next_char = chars.shift()
chunksize += next_char.bytesize
chunks << next_char
end
end
message_chunks << chunks.join
end
client = Ably::Rest.new(ENV["ABLY_PRIVATE_KEY"])
channel = client.channel(channel_name)
message_id = SecureRandom.uuid
message_chunks.each_with_index do |chunk, index|
data = {
utf: "✓",
message_id: message_id,
message_index: index,
message_count: message_chunks.count,
data: chunk
}
channel.publish(event, data)
endRebuliding on client side:
let incomingMessages = {}
const handleMessageChunk = (type, messageChunk) => {
if (messageChunk.message_id && messageChunk.message_count && (messageChunk.message_index || messageChunk.message_index === 0)) {
if(incomingMessages[messageChunk.message_id] === undefined) {
incomingMessages[messageChunk.message_id] = {}
}
incomingMessages[messageChunk.message_id][messageChunk.message_index] = messageChunk
let isMessageComplete = Object.keys(incomingMessages[messageChunk.message_id]).length == messageChunk.message_count
if (isMessageComplete) {
let completeMessage = ""
for (let i = 0; i < messageChunk.message_count; i++) {
completeMessage += incomingMessages[messageChunk.message_id][i].d;
}
delete incomingMessages[messageChunk.message_id]
console.log("message completed", JSON.parse(completeMessage), incomingMessages)
dispatch({type: type, payload: JSON.parse(completeMessage)})
} else {
//console.log("No finished message", messageChunk)
}
}
}┆Issue is synchronized with this Jira Bug by Unito
Right now we only get an exception, so we built a splitting mechanism by ourselfes, but it would be great if your lib would adjust the message size automatically. Also sometimes we are still hitting the limits (even though we are calculating all the bytes by ourselfes), so maybe you have a better solution?
This is our solution (we are rebuilding the message on the client side afterwards):
Rebuliding on client side:
┆Issue is synchronized with this Jira Bug by Unito