hackerspacesg / hackerspace.sg

Hackerspace.sg 2.0 using Punch
https://hackerspace.sg/
17 stars 25 forks source link

readded fallback to normal PNG, sans retina which gets loaded along with... #111

Closed valentine closed 9 years ago

valentine commented 9 years ago

... SVG

for #107

kaihendry commented 9 years ago

Please provide quantitative evidence that at least 1% of our website sessions will use this fallback?

valentine commented 9 years ago

Fallbacks are for edge cases. This one doesn't affect performance since the PNG doesn't load if the SVG does.

kaihendry commented 9 years ago

I don't understand that code but it seems to work in the sense the PNG doesn't get loaded.

rolandturner commented 9 years ago

Very funny.

On 11/11/2014 06:06 PM, Kai Hendry wrote:

Please provide //quantitative// evidence that at least 1% of our website sessions will use this fallback?

— Reply to this email directly or view it on GitHub https://github.com/hackerspacesg/hackerspace.sg/pull/111#issuecomment-62526280.

robertsky commented 9 years ago

Edge case is really edge, like ie8 and below and partial support for Android browser 4.3 and below (no masking). http://caniuse.com/#feat=svg Looking at GA for the past three months, 113 or 1.67% of 6769 sessions are Andriod browser 4.2 and below.

LOL.

rolandturner commented 9 years ago

(N.B. broadened to list)

This discussion points to a broader need to establish a consensus on tolerable side-effect threshold, particularly for improvements that are motivated by nothing more than (understandable!) irritation at the endless need to accommodate non-awesome software used by other people (vs. say some material improvement in HackerspaceSG's ability to meet its organisational goals which, presumably, is the purpose of the website).

I'd suggest that a bare minimum threshold is that we never remove a working feature or collection of related working features (or indeed make any change or collection of related changes) which would cause appreciably negative impact for 1% of sessions from the preceding 30 days. I'd suggest that there is reason to consider tightening both of those thresholds, but that I'm more interested in the existence of an consensus threshold to prevent further irritation-driven harm to the website than I am about the specifics:

Kai may be thrilled to observe an interesting side-effect of drawing the line this way: a rationale is brought about for removing pages from the main body of the website which few people visit. This could perhaps be addressed by creating a Miscellany top-level section to move unpopular pages to, which would help solve two additional problems:

Comments welcome.

On 11/11/2014 09:54 PM, robertsky wrote:

Edge case is really edge, like ie8 and below and partial support for Android browser 4.3 and below (no masking). http://caniuse.com/#feat=svg Looking at GA for the past three months, 113 or 1.67% of 6769 sessions are Andriod browser 4.2 and below.

LOL.

— Reply to this email directly or view it on GitHub https://github.com/hackerspacesg/hackerspace.sg/pull/111#issuecomment-62549199.

valentine commented 9 years ago
  • Valentine has suggested 90 days instead of 30 days. I think that this is a pretty good idea but, as above, am more interested in establishing consensus then in where it lands, specifically.

I don't recall suggesting this.

The fallback was earlier removed because the PNG was downloaded even though the SVG was the only thing being rendered. This led to #109, where the PNG was removed altogether.

I reintroduced the fallback because it provides no detriment to users of current browsers, and is in line with our goals of involving as many people from different walks of life in the world of tech. I'm not sure if we should be selective of our visitors — it may be valid for other companies, or for non-techies looking to use HSG as a co-working space, but it doesn't provide a favourable impression nonetheless.

The idea for a Miscellany section is good. I've always wanted to separate the operational content from those that a new visitor would like to see, and to introduce tables of contents where appropriate (taking a cue from Wikipedia). It's not very impatient-newbie-friendly at the moment.

rolandturner commented 9 years ago

On 11/13/2014 05:51 PM, Valentine C wrote:

  * Valentine has suggested 90 days instead of 30 days. I think
    that this is a pretty good idea but, as above, am more
    interested in establishing consensus then in where it lands,
    specifically.

I don't recall suggesting this.

My mistake. It appears that it was icedwater (Kheng Hui?):

That is only over the last month, which is why I asked for more data. Intuitively, the trend should be the same, but I don't think one month is enough to be safe.

  • Roland