USCiLab
Splitting / Combining Archives?
Hello,
I was wondering if it's possible to have one archive split into multiple smaller archives? Specifically, what I'm trying to do is serialize a lot of polymorphic pointers, but I'd like every pointer entry to be serialized in its own output file. Then later on when deserializing I'd like to load all those archives, smash them together, and have the pointers re-initialize themselves properly.
The end result would be that if I 1/10000 pointers changes, the whole archive doesn't have to be regenerated again, but rather, just one archive file for that specific pointer would be remade.
You’re likely to have better luck generating some kind of paginated archive class which contains an array of “pages” each “page” object contains a list of polymorphic objects.
Have manual load/save methods which call the archive load/save code and rebuild the pointers after loading.
This isn’t something I think you can expect to be built into cereal itself though (at least that’s my gut reaction based on your description.)
On Tue, Dec 31, 2024 at 9:13 PM Damir Halilovic @.***> wrote:
Hello,
I was wondering if it's possible to have one archive split into multiple smaller archives? Specifically, what I'm trying to do is serialize a lot of polymorphic pointers, but I'd like every pointer entry to be serialized in its own output file. Then later on when deserializing I'd like to load all those archives, smash them together, and have the pointers re-initialize themselves properly.
The end result would be that if I 1/10000 pointers changes, the whole archive doesn't have to be regenerated again, but rather, just one archive file for that specific pointer would be remade.
— Reply to this email directly, view it on GitHub https://github.com/USCiLab/cereal/issues/849, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAHBPZMPDXPOXQLHPEOH4S32INMMXAVCNFSM6AAAAABUOF3DJCVHI2DSMVQWIX3LMV43ASLTON2WKOZSG43DIOBQG43TOOA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
I suppose all I would have to do is defer the pointer rebuilding after all individual archives were loaded and combined into what cereal expects potentially.