Bugzilla serves search results as
multipart/x-mixed-replace, a mime-type which lets it send the “Please stand by…” page, have that displayed, and then replace it with the search results once the query completes. That’s probably fairly tolerable if you have a fast connection, a slow Bugzilla, and typically small queries. Me, I don’t.
I tend to do triage, when I do it, with broad “Find a Specific Bug” queries including closed bugs, to pull in all the synonyms that appear in duplicates, and sometimes (for narrow enough components, like Password Manager) with bookmarked searches for every open bug in a component, and I’m doing that over dialup. The “Please stand by…” splashscreen is tolerable enough, but then as my 200 results trickle in, and I scroll down them and find the bug I was after while the page is still loading, I start to get the second set of replaced content, so that my click on the link doesn’t actually load the bug I want, it just hurries along the re-loading of the same list that had already served its purpose. That’s particularly annoying when I click a bug link, switch to another tab (to do something like start the process of marking a duplicate) only to come back and find that the tab that should have my dupe target still has the results list that has my target in it. Somewhere.
Luckily, the other day I was reading buglist.cgi and discovered that I can quite often avoid the
multipart/x-mixed-replace, because any URL that includes
&serverpush=0 just gets served straight
text/html. Fixing the search forms would require a bookmarklet, or more usefully a Greasemonkey script, to insert a
<input type="hidden" name="serverpush" value="0"> in the form, but for my main bookmark and quicksearch use, just adding it to the URL makes using Bugzilla vastly more pleasant. Of course, I really ought to fix bug 298995 to get a user pref to turn it off completely, but now that my itch is scratched…