Re: [PATCH] nicer eye candies for pack-objects

I like this, but like the "every second or every percent
whichever comes first" unpack-objects does even better.  How
about something like this on top of your patch?

---

diff --git a/pack-objects.c b/pack-objects.c
index 5e1e14c..d05ab23 100644
--- a/pack-objects.c
+++ b/pack-objects.c

@@ -674,10 +674,14 @@ static void find_deltas(struct object_en
 	int i, idx;
 	unsigned int array_size = window * sizeof(struct unpacked);
 	struct unpacked *array = xmalloc(array_size);
+	unsigned processed = 0;
+	unsigned last_percent = 999;
 
 	memset(array, 0, array_size);
 	i = nr_objects;
 	idx = 0;
+	if (progress)
+		fprintf(stderr, "Deltifying %d objects.\n", nr_objects);
 
 	while (--i >= 0) {
 		struct object_entry *entry = list[i];

@@ -686,10 +690,15 @@ static void find_deltas(struct object_en
 		char type[10];
 		int j;
 
-		if (progress_update || i == 0) {
-			fprintf(stderr, "Deltifying (%d %d%%)\r",
-				nr_objects-i, (nr_objects-i) * 100/nr_objects);
-			progress_update = 0;
+		processed++;
+		if (progress) {
+			unsigned percent = processed * 100 / nr_objects;
+			if (percent != last_percent || progress_update) {
+				fprintf(stderr, "%4u%% (%u/%u) done\r",
+					percent, processed, nr_objects);
+				progress_update = 0;
+				last_percent = percent;
+			}
 		}
 
 		if (entry->delta)

On Wed, 22 Feb 2006, Junio C Hamano wrote:

I like this, but like the "every second or every percent
whichever comes first" unpack-objects does even better.  How
about something like this on top of your patch?

Well... my concern is (if I'm right) that this status is generated 
remotely and sent over the network when performing a fetch.  The "every 
percent" might in this case generate quite some significant overhead if 
the pack is small.

Also (personal opinion) such progress numbers are harder to read when 
they change too fast.


Nicolas

On Wed, 22 Feb 2006, Nicolas Pitre wrote:

On Wed, 22 Feb 2006, Junio C Hamano wrote:

quoted

I like this, but like the "every second or every percent
whichever comes first" unpack-objects does even better.  How
about something like this on top of your patch?

Well... my concern is (if I'm right) that this status is generated 
remotely and sent over the network when performing a fetch.  The "every 
percent" might in this case generate quite some significant overhead if 
the pack is small.

Well, my thinking behind the original unpack-objects behaviour was that we 
don't really care about the max 100 extra packets. 

The days of 150 baud line printers are gone. I cannot imagine any valid 
situation where you can't send a hundred small packets to update the 
screen.. And it did make a nice visual difference.

(In fact, over a network, if the line really is slow, you'll find that 
Nagle will fix things, and you'll just see one extra - but obviously 
slightly bigger - packet).

		Linus

Nicolas Pitre wrote:

On Wed, 22 Feb 2006, Junio C Hamano wrote:

quoted

I like this, but like the "every second or every percent
whichever comes first" unpack-objects does even better.  How
about something like this on top of your patch?

Well... my concern is (if I'm right) that this status is generated 
remotely and sent over the network when performing a fetch.  The "every 
percent" might in this case generate quite some significant overhead if 
the pack is small.

But if the pack is small it won't matter. It's when it's big we want to 
know about it (and then the each-percent method is better, really).

Also (personal opinion) such progress numbers are harder to read when 
they change too fast.

I don't know about the rest of the world, but when I see numbers 
counting up with a percent-sign behind them I do some mental math to see 
how many brain-ticks go between each increment and then multiply with 
100 to see if I need to get a beer while waiting. I don't really care 
what number it shows if it flashes too fast to read.

-- 
Andreas Ericsson                   andreas.ericsson@op5.se
OP5 AB                             www.op5.se
Tel: +46 8-230225                  Fax: +46 8-230231

On Wed, 22 Feb 2006, Linus Torvalds wrote:

Well, my thinking behind the original unpack-objects behaviour was that we 
don't really care about the max 100 extra packets. 

Obviously, the "every percent" has at most 100 additional packets.

And if it updates too fast to be readable, that means it'll be over in a 
snap anyway.

So I don't have any objections left.


Nicolas