wal write of large object

<div dir="ltr">Hi, hackers<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">I&#39;m eager to dive into how to write wal for large object. In the code path:<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">heap_insert -&gt; heap_prepare_insert -&gt; heap_toast_insert_or_update -&gt; toast_save_datum -&gt; heap_insert<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">I find heap_insert is called back.<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">1. At heaptup = heap_prepare_insert(relation, tup, xid, cid, options), the comment says tup is untoasted data, but it could come from toast_save_datum, still untoasted ?<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">2. At toast_save_datum while loop, we know heap_insert is called with every chunk data, so every chunk data is written to WAL by XLogInsert() , right ? Although it may make WAL big. (It is still the case regarding to blob, cblob ?)<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">3. When heap_insert is called firstly and heap_prepare_insert returns, what does heaptup mean for large object which has been chunked and written by toast_save_datum ?<br>
</div><div dir="ltr"><br>
</div><div dir="ltr"><br>
</div><div dir="ltr">Any articles about this aspect would be appreciated.<br>
</div><div dir="ltr"><br>
</div>