Open beaufortfrancois opened 7 years ago
Wouldn't that concern "not fetch the whole content at once into renderer memory like arrayBuffer does." a Blink only thing?
So far down the line, may be too late to add support for blob...
IIRC at some point there was the intention of having appendBuffer taking a ReadableStream which would allow not loading the whole buffer in memory. I just checked the status of this and could not find it anywhere.
I believe you mean appendStream... it was removed in #139
Can we close this issue as a duplicate of ISSUE-14?
/paulc HME WG Chair
@paulbrucecotton Do we have a rough ETA when to expect the next version of MSE (including appendStream()
)?
The HME WG's charter currently only permits the WG to do maintenance on MSE. In Philippe's announcement it clearly states:
"As such, the Working Group is only allowed to make corrections that do not add new features [4] to its Recommendation(s) under this charter extension."
My suggestion is that those interested in the appendStream feature should "incubate" such a feature in a Community Group.
/paulc HME WG Chair
Thank you @paulbrucecotton @wolenetz Is that something you would be interested in?
+1
@calvaris
IIRC at some point there was the intention of having appendBuffer taking a ReadableStream which would allow not loading the whole buffer in memory
Rather than take a readable stream, it should expose a writable endpoint that you could pipe to.
I threw together a little demo of this at https://media-source-writable.glitch.me/, based on an example from @beaufortfrancois.
See also #14, probably a duplicate if the approach is to add a writable endpoint to a SourceBuffer to which a stream/sequence of chunks could be piped.
Is there's a technical reason why blob is not a valid source for
appendBuffer()
? The immediate benefit I see to use blob is to not fetch the whole content at once into renderer memory like arrayBuffer does. It would be especially true in responses from the Cache API.