pointers suggestion

new topic     » topic index » view thread      » older message » newer message

Or: Re: Andy Serpa's memory problem.

If Eu had pointers we could optionally use with variable_id, when we knew we 
were going to grow several sequences at a time, one element at a time, for 
several megabytes, and then destroy them and recreate them, repeatedly (in 
other words, we knew what we were doing, and had to do it this way 
anyhow), we could pull an olde pascal trick (yup, i been looking at my TP 
code for dos again):

<pseudocode>
location1 = allocate(8388608)
location2 = allocate(8388608)
location3 = allocate(8388608)
-- location* now holds 8megabytes for themselves
-- and they together (possibly) occupy a single block of 24megs

sequence var1, var2, var3
fix var1 = location1
fix var2 = location2
fix var3 = location3

Now, Bach/Bliss/Eu can mangle the heck out of var* and not run into 
garbage collection lag, memory fraging, repeated de/allocations as the var*s 
grow and collapse, and best of all we know where the vars live. Given how Eu 
already handles resizing automagically, outgrowing the allocated space 
could result in simply falling thru to the innate allocation code to grow it 
bigger, but not smaller.

Kat,
still using goto
still using string execution

new topic     » topic index » view thread      » older message » newer message

Search



Quick Links

User menu

Not signed in.

Misc Menu