Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Chrome bloat brainstorm (groups.google.com)
97 points by crizCraig on March 22, 2011 | hide | past | favorite | 23 comments


The answer to "why" given in that thread is very interesting:

"Many reasons, but two off the top of my head: 1. We do distribution deals with Chrome, where we bundle Chrome with other products. These get difficult when our binary grows. 2. We see increased download failures / install dropoffs as the binary grows, especially in countries with poor bandwidth like India. India also happens to be a very good market for Chrome (we have good market share there and growing), so that's also very problematic."


This is a beautiful thing in a world of 600mb printer driver install packages.


I was recommended a new mapping program for my iphone yesterday, and when I went to download it, it said that I needed 4.2GB of space to install.

Also as a "haha, crappy bandwith!" aside: I found it strange to think of most people only having -700kbps in the states. I have a 100mbps line for 50USD/month and regularly get around 60mbps. I have to install chrome on various computers for work, and I've never even noticed more than a few second delay between downloading the binaries and installation.


Not all Americans can afford $50 USD / month for a luxury service. There are also wide swaths of the US that don't have any better options since telecoms refuse to build more infrastructure.


Agreed. My company's offering opened my eyes to two things:

1. Many people, even in the States, have sloooow connections at home. 2. Many people have very poor cell/GSM/GPRS coverage. We still have to provide a POTS modem for our handheld data collection devices[1].

[1] We just released a new device (smartphone) that supports a POTS modem. Think of that! A cellphone that transfers data via a dial up modem.


$50/month!? Luxury service!? We pay $50/month for 768KBps (6Mbps), and that's the second cheapest service in town. I'm going to go cry now.

P.S. Mapping software is by nature huge. It typically contains, you know, maps. Of the whole country.


For an iPhone app though? Meant to be installed over the air? That's incredibly poorly thought out.

Better would be to allow you to download which maps you need as you need them, to shrink the initial installer down. Better than that would be having the maps be online accessible, and are immediately cached to the device after first use. Best is probably somewhere between those two, or something along those lines, but I can't imagine who packaged up a 4.2Gb iphone app and didn't ask any questions about that.


It's probably more robust to let iTunes download the whole binary, instead of have the user attempt to download specific maps over WiFi (the device may go to sleep, run out of battery, etc).


Come on now. Surely resuming a partial download is a solved problem in this day and age?


Funny how I tried to download iOS 4.3 from iTunes for my new iPhone 4 yesterday and the computer went to sleep while downloading it, which is around 700Mb. When I noticed I woke it up and clicked on resume, bang, went from 400Mb to "zero Kb" right away.

And, uh, it happened twice... Finally, at my third attempt I was able to download the whole thing after deactivating automatically sleep mode. As soon as it finished, an error was shown and so the file was lost again.


I still find iOS app updates on the device a bit hit and miss.


You are quite privileged. I pay $60 per month for one tenth of that in San Diego (not exactly hillbilly country).


I was able to get great connection speeds in San Diego with Time Warner for a little less than $60 a month - if you upgrade to their turbo package just make sure that they give you a compatible router; I spent five months paying for the better service without being able to access it because they didn't upgrade the hardware.


I wonder if it's possible to have a basic browser that is very small and has most of the functionality with minimum 3rd party libraries like Flash/PDF plugin. Once the browser starts, the full software installation continues, just like Chrome's silent software update. Slowly, the users have the entire browser installed!


Yeah, I tried to post a similar suggestion on the group. Stuff like developer tools, native client, gpu, themes, and extensions might be able to get lazy loaded.


@Software devs: How are modules implemented in softwares that you create? Are dlopen()s sufficient or are there issues?


I hear there's a team at Adobe that everyone loves to hate that consistently adds features and keeps binary sizes down year after year...

True story.


Here are the historical file sizes of Adobe's Flash Player installer (compressed .exe for Windows ActiveX Control):

  * Flash Player  2   (1997) = 0.2 MB
  * Flash Player  3   (1998) = 0.2 MB
  * Flash Player  4   (1999) = 0.3 MB
  * Flash Player  5   (2000) = 0.3 MB
  * Flash Player  6   (2002) = 0.5 MB
  * Flash Player  7   (2003) = 0.6 MB
  * Flash Player  8   (2005) = 0.8 MB
  * Flash Player  9   (2006) = 1.5 MB
  * Flash Player 10   (2008) = 1.9 MB
  * Flash Player 10.1 (2010) = 2.7 MB
  * Flash Player 10.2 (2011) = 2.7 MB


and I heard there's a team that doesn't add that icon you deleted back to your desktop every time an update is released.


Netscape 2.02 was 3.13 megs, for comparison.


Huh, I hadn't realized Netscape was that big way back in 1996! From that point of reference, Chrome's 26.2 megs doesn't actually seem huge. If we take 384 kbps DSL to be today's equivalent to a 14.4 kbps modem (low end of the "normal" internet connection), it's a 3x quicker download as well.



There's not really much more information in there beyond what's in the google groups thread. Pretty good summary though.




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: