Bug Triage

[Jeff Kaufman]

I just went over the last few months of bugs, adding things to
https://github.com/pagespeed/mod_pagespeed/wiki/Work-Prioritization
Here are the bugs that people aren't already working on that would be
good to handle soon. CSP support is the big one, but we knew that.
Most of the others should be relatively contained short fixes.

* content security policies
https://github.com/pagespeed/mod_pagespeed/issues/876

* ipro messes with application/json headers
https://github.com/pagespeed/mod_pagespeed/issues/1321

* don't mangle files that start with gzip magic bytes
https://github.com/pagespeed/mod_pagespeed/issues/1307

* respect s-maxage and age
https://github.com/pagespeed/mod_pagespeed/issues/1287

* rpms are missing fortify_source
https://github.com/pagespeed/mod_pagespeed/issues/1301

* fetching pages with +debug enables other filters
https://github.com/pagespeed/ngx_pagespeed/issues/1190

* inline_google_font_css should use minification
https://github.com/pagespeed/ngx_pagespeed/issues/1183

* dedup_inlined_images fails if you switch away from the page mid-load
https://github.com/pagespeed/mod_pagespeed/issues/1002

* property cache lifetime should be configurable
https://github.com/pagespeed/mod_pagespeed/issues/1311

* disallow and allow don't interact right with merging
https://github.com/pagespeed/mod_pagespeed/issues/1294

* Allow disabling optimizations via tags on attributes
https://github.com/pagespeed/mod_pagespeed/issues/1320

Jeff

[Steve Hill]

I looked over these. Many of them look like pretty straightforward fixes and I think we should try to fix them for the next release.

The disallow/allow thing is on me to either repro or close, since Josh couldn't repro and also added a test that it actually does DTRT. The attributes thing sounds non-trivial to me since you'd have to dynamically generate a new RewriteOptions pretty late in the game, but maybe it's not that bad.

[Jeff Kaufman]

On Wed, Jun 29, 2016 at 3:46 PM, Steve Hill <che...@google.com> wrote:
> I looked over these. Many of them look like pretty straightforward fixes and
> I think we should try to fix them for the next release.

Great!

>
> The disallow/allow thing is on me to either repro or close, since Josh
> couldn't repro and also added a test that it actually does DTRT.

I reproed it, and added a comment:
https://github.com/pagespeed/mod_pagespeed/issues/1294

> The
> attributes thing sounds non-trivial to me since you'd have to dynamically
> generate a new RewriteOptions pretty late in the game, but maybe it's not
> that bad.

Maybe it should only be disallow, and should hide the element from the
named filters?

Jeff

[Steve Hill]

Looking down this list, most of the simple stuff has been fixed and marked for backporting in the spreadsheet. One notable exception is #1301 (FORTIFY_SOURCE) which is fixed but not backported. Do we feel safe backporting that?

Of this list, the only one I think that's worth considering for a fix is #1190 (+debug enables other filters) but I suspect that might be hard.

Is there anything else that really should go into the old branch?

[Steve Hill]

Sorry, I should add that Jeff's cache cleaning stuff is probably good to go in too.

[Jeff Kaufman]

I think #1190 was fixed by
https://github.com/pagespeed/mod_pagespeed/commit/4f657b92984c19ea562b2279cb2ff3a634a2a1ca
; someone just needs to check.

On Mon, Jul 18, 2016 at 11:39 AM, Steve Hill <che...@google.com> wrote:

[Otto van der Schaaf]

Just checked, nps  #1190 was not fixed:

https://github.com/pagespeed/ngx_pagespeed/issues/1190#issuecomment-233596954

--
You received this message because you are subscribed to the Google Groups "pagespeed-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pagespeed-de...@googlegroups.com.
To post to this group, send email to pagesp...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pagespeed-dev/CAMJ6YUvrBSNE1KE6aD%2BDresZj0CVXVxzheNTzi8N8X-Ln%3DCrVw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Jeff Kaufman]

The (draft fix) ended up being pretty safe and I think we should
probably backport it:
https://github.com/pagespeed/ngx_pagespeed/pull/1228