Compiler requirements #22
Conversation
|
QQ, do you have an ETA for a draft of this? |
Add first draft
|
General feedback: overly focused on implementation details, not enough on the impact of the situation. It's hard to reason about ROI from what is provided at time of writing. |
|
|
||
| ## What Do We Need? | ||
|
|
||
| An at least 10x faster compiler using less memory. |
There was a problem hiding this comment.
It takes 30 minutes to compile Noto therefore 10x faster doesn't follow.
How often do people compile Noto? What happens if it takes half an hour? Why is there ROI for Google in making take half the time vs a tenth vs a hundredth, where does the return fall off?
|
|
||
| As soon as possible. | ||
|
|
||
| ## Implementation Requirements |
There was a problem hiding this comment.
At a glance none of these appear to be requirements. Recommendations might be more appropriate.
|
|
||
| ## When Do We Need It? | ||
|
|
||
| As soon as possible. |
There was a problem hiding this comment.
Not helpful. More useful would be some description of the impact of not having a faster compiler over different periods of time. If there is no speedup for a year, what harm would there be, how much does it cost us?
What if we could make compiling Noto take 10 minutes through a modest investment in improving fontmake and achieving 10+x faster was a large investment, which way should we go and why?
|
|
||
| [how do these diverge from those required for subsetting?] |
There was a problem hiding this comment.
subsetting is typically done on a font binary
Closes #19